regional mortality variation in finland: a study of two

43
GENUS, LXII (No. 2), 169-211 169 JAN SAARELA – FJALAR FINNÄS Regional mortality variation in Finland: a study of two population groups 1. INTRODUCTION From a demographic and public health perspective, the two native population groups in Finland constitute interesting objects of study. Several analyses have shown that members of the Swedish-speaking minority have lower death risks than Finnish speakers at all ages (Sauli, 1979; Valkonen et al., 1992; Koskinen, 1994; Martelin, 1994). What makes comparisons so challenging is that both groups are indigenous, in most observable respects equal, and live under similar environmental circumstances. Like in some other European countries (Langford and Bentham, 1996; Luy, 2004), in Finland too there is a striking variation in mortality across regions. Death rates at all ages have a tendency to increase gradually in the South-West to North-East direction (Koskinen and Martelin, 1994; Pitkänen et al., 2000), and these relative differences between regions have remained fairly constant during the past decades (Statistics Finland, 1991; 1997). In spite of the academic and public attention directed towards the issues, no exhaustive explanation has yet been provided. The mortality differential between Finnish speakers and Swedish speakers, as well as the regional variation in death rates, can only partly be attributed to socio- demographic and other structural conditions. As a consequence, it has been argued that latent determinants should be sought in an array of factors. Principally, one may distinguish four potential phases during which the differentials have arisen: at conception through genes, in-utero, during childhood and adolescence, or at adult ages through environmental and thus also behavioural factors (cf. Doblhammer, 2004). In relation to the data and the approach for analysis in this paper, we discuss mainly the potential importance of genes and conditions in early life, as opposed to circumstances that may be of relevance in adult life. It should be borne in mind, though, that genes are individual-related and time consistent, and thus have a biological interpretation, whereas environmental conditions are contextual and therefore tend to have a more ecological interpretation. Koskinen and Martelin (2003) have found that the Finnish-Swedish mortality differential in working ages is widest in cardiovascular diseases. In their opinion, that result indicates that genetic factors could be important.

Upload: others

Post on 27-Apr-2022

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Regional mortality variation in Finland: a study of two

GENUS, LXII (No. 2), 169-211

169

JAN SAARELA – FJALAR FINNÄS

Regional mortality variation in Finland: a study of two population groups

1. INTRODUCTION

From a demographic and public health perspective, the two native

population groups in Finland constitute interesting objects of study. Several analyses have shown that members of the Swedish-speaking minority have lower death risks than Finnish speakers at all ages (Sauli, 1979; Valkonen et al., 1992; Koskinen, 1994; Martelin, 1994). What makes comparisons so challenging is that both groups are indigenous, in most observable respects equal, and live under similar environmental circumstances.

Like in some other European countries (Langford and Bentham, 1996; Luy, 2004), in Finland too there is a striking variation in mortality across regions. Death rates at all ages have a tendency to increase gradually in the South-West to North-East direction (Koskinen and Martelin, 1994; Pitkänen et al., 2000), and these relative differences between regions have remained fairly constant during the past decades (Statistics Finland, 1991; 1997).

In spite of the academic and public attention directed towards the issues, no exhaustive explanation has yet been provided. The mortality differential between Finnish speakers and Swedish speakers, as well as the regional variation in death rates, can only partly be attributed to socio-demographic and other structural conditions. As a consequence, it has been argued that latent determinants should be sought in an array of factors. Principally, one may distinguish four potential phases during which the differentials have arisen: at conception through genes, in-utero, during childhood and adolescence, or at adult ages through environmental and thus also behavioural factors (cf. Doblhammer, 2004). In relation to the data and the approach for analysis in this paper, we discuss mainly the potential importance of genes and conditions in early life, as opposed to circumstances that may be of relevance in adult life. It should be borne in mind, though, that genes are individual-related and time consistent, and thus have a biological interpretation, whereas environmental conditions are contextual and therefore tend to have a more ecological interpretation.

Koskinen and Martelin (2003) have found that the Finnish-Swedish mortality differential in working ages is widest in cardiovascular diseases. In their opinion, that result indicates that genetic factors could be important.

Page 2: Regional mortality variation in Finland: a study of two

JAN SAARELA – FJALAR FINNÄS

170

Some other indirect support for the genetic explanation is that regional mortality variation at the national level can be attributed to a great extent to coronary diseases (Koskinen and Martelin, 1998), and that mortality by ischaemic heart disease have been found to be more strongly correlated with people’s birth region than with their region of residence (Valkonen, 1987). In the event that genetic variation is associated with varying death rates, we believe, however, that it acts foremost via hereditary diseases at higher ages.

There are also some indirect evidence in support of differences in lifestyles between Swedish speakers and Finnish speakers. In working ages, there is a higher prevalence of male suicides and accidents related to violence in Finnish speakers (Koskinen and Martelin, 2003), and a twice higher risk of alcohol-related mortality (Blomgren et al., 2004). Relatively more Finnish speakers than Swedish speakers have also been found in socially disadvantaged subgroups with elevated mortality risks (Saarela and Finnäs, 2005a).

Working-aged Finnish speakers, who live in the coastal area of Finland but originate from the eastern and northern parts of the country, have been shown to have higher mortality levels than other people in the coastal area (Saarela and Finnäs, 2005b). On the basis of those findings it is still difficult to say how much of the mortality differential is related to lifestyles and how much to genetics and early-life conditions, because latent factors are not necessarily the same in all ages. Mortality caused by unhealthy behaviour and risky ways of living is relatively more common in younger people than in older ones, and it is not before the age of 50 that deaths caused by diagnosed diseases start to dominate (Statistics Finland, 2005a). When considering how causes of death are distributed across age groups, one may therefore expect a more direct interrelation between lifestyles and death risks in younger people, whereas general health conditions would matter more at higher ages.

Further, one cannot rule out the possibility that variation in socio-economic and other living conditions in early life underlie the difference in mortality between the population groups. It is fairly clear, however, that present socio-economic circumstances cannot explain the mortality differential (Saarela and Finnäs, 2005a).

A methodological difficulty with all the explanations discussed is that there are no adequate registers that would provide explicit information about the issues. Still, we are convinced that current register data can be used to shed additional light on the contribution of the proposed determinants. In this paper, this is accomplished by focusing on people aged 60 years and over. In this age group the number of deaths is large enough to facilitate a rigorous study, and a great part of all deaths is obviously related to the

Page 3: Regional mortality variation in Finland: a study of two

REGIONAL MORTALITY VARIATION IN FINLAND…

171

individual’s overall health status. In our analysis, we include birth region to reflect circumstances operating before or in early life. Thus we aim at exploring the Finnish-Swedish mortality differential in terms of accounting for regional mortality variation and specifically an individual’s region of birth. Our approach can therefore, principally, be related to previous studies from other countries, which have looked at internal migrants and found that a person’s birth region is an important determinant of mortality (Strachan et al., 1995; Schneider et al., 1997a; 1997b; Regidor et al., 2002; Doblhammer, 2004). Considering that Finnish speakers and Swedish speakers differ in geographical distribution within the country, it is necessary to put specific focus on the regions where both the population groups live.

Our statistical analyses are based on a longitudinal population census register compiled by Statistics Finland. It covers a 25-year period from 1970 to 1995. All cohorts under study were born before the large population shifts in Finland took place. At birth, they therefore represent a geographically stable population. Birth region should consequently be considered a reasonable indicator of a person’s geographical extraction.

The data are well suited for studying relative differences in adjusted death risks as they contain individual-level information about basic socio-demographic factors for the whole population in Finland. They are not equally appropriate to describe overall mortality development, however, or to quantify the Finnish-Swedish mortality differential in a conventional manner, because the censuses have been undertaken at five-year intervals. We therefore start with presenting some basic results of calculations based on ordinary vital statistics for all ages. This can be motivated also by the fact that, in spite of a great variety of studies concerning the Finnish-Swedish mortality differential, no thorough documentation yet exists.

2. MORTALITY DEVELOPMENT IN FINLAND Since the late 1970s publications on vital statistics have contained

information about the annual number of deaths and mean populations by age and sex for both the total population of Finland and for the Swedish speakers. We have used these data to calculate age-specific death rates and life expectancies for Finnish speakers and for Swedish speakers. The highest closed age intervals available are 80-84 for initial years and 85-89 for more recent years. Calculations in the very old are therefore based on approximations and extrapolations according to published life tables for the corresponding years in Finland’s population.

Page 4: Regional mortality variation in Finland: a study of two

JAN SAARELA – FJALAR FINNÄS

172

All data are for five-year age intervals. To simplify presentation we combine these into broader intervals and standardise for variation in age distribution across years and population groups, using Finland’s population in 2002 as the standard population. Death rates still vary considerably over time for Swedish speakers, as they are fairly small in number. Firstly, we therefore discuss general mortality development based on the results for Finnish speakers.

Age-specific mortality trends are presented for men in Figure 1 and for women in Figure 2. They show that there has been a large decline in overall death rates since the late 1970s, which corresponds to the situation in most industrialised countries. The mortality decline tends to have been smaller in higher age groups than in younger ones, with the exception that people aged 20-39 years experienced no reduction in mortality risks during the 1980s. Still, it is important to realise that the description refers to relative changes in death rates. As overall death rates increase exponentially by age, the decrease in absolute numbers is by far the largest in older age groups.

In our opinion, the overall picture still is that there has been a fairly continuous mortality decline in all ages. Death rates have decreased by some 40 per cent since the late 1970s. In terms of life expectancy for men this corresponds to an increase of 6.6 years until 2003, from 68.5 years to 75.1 years, and for women with an increase of 4.0 years, from 77.8 to 81.8 years (Statistics Finland, 2004).

In spite of the seemingly large mortality decline in younger ages, their impact on life expectancy is rather small, because the total number of deaths in these ages is low. For instance, changes in the age span 0-59 years contribute to only 0.4 years of the total increase of 4.0 years in female life expectancy, and to 2.2 years of the total increase of 6.6 years in male life expectancy.

Corresponding rates were calculated in a similar manner for the Swedish speakers. To facilitate comparisons we outline these death rates in relation to those of Finnish speakers. To eliminate some random variation the curves are smoothed by a five-point weighted moving average. Figure 3 is for men and Figure 4 for women.

Mortality development has been very similar in both population groups. The Finnish-speaking excess mortality has consequently not diminished over time. In ages over 40 years, the death rates of Swedish-speaking women are about 10 per cent lower than those of Finnish-speaking women. In men aged over 75 years, the relative advantage of Swedish speakers is at this same level, but even higher in younger ages.

Life table estimates were calculated to quantify these differences in terms of life expectancy at different ages. The results are presented in Table

Page 5: Regional mortality variation in Finland: a study of two

REGIONAL MORTALITY VARIATION IN FINLAND…

173

Figure 1 – Development of the age-specific death rate for men in Finland by age group

0,3

0,4

0,5

0,6

0,7

0,8

0,9

1

1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002

0-19 years

20-39 years

40-59 years

60-74 years

75-89 years

Note: 1978=1.

Figure 2 – Development of the age-specific death rate for women in Finland by age group

0,3

0,4

0,5

0,6

0,7

0,8

0,9

1

1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002

0-19 years

20-39 years

40-59 years

60-74 years

75-89 years

Note: 1978=1.

Page 6: Regional mortality variation in Finland: a study of two

JAN SAARELA – FJALAR FINNÄS

174

Figure 3 – Age-specific death rate of male Swedish speakers in relation to that of male Finnish speakers by age group

0,5

0,6

0,7

0,8

0,9

1

1,1

1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000

20-39 years

40-59 years

60-74 years

75-89 years

Note: five-point weighted moving average. Figure 4 – Age-specific death rate of female Swedish speakers in relation to

that of female Finnish speakers by age group

0,5

0,6

0,7

0,8

0,9

1

1,1

1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000

20-39 years

40-59 years

60-74 years

75-89 years

Note: five-point weighted moving average.

Page 7: Regional mortality variation in Finland: a study of two

REGIONAL MORTALITY VARIATION IN FINLAND…

175

1, where data have been combined into four time periods. Calculations for the highest age groups should be interpreted with some caution as they are based on extrapolations. The difference in life expectancy at birth between Swedish speakers and Finnish speakers is found to be about three years for men and just over one year for women. As there is no evident time variation in the Finnish-Swedish mortality differential, the difference in life expectancy between the two population groups has also remained fairly constant. Based on the above findings we estimate that, in 2003, life expectancy for Swedish-speaking men was just over 78 years, and about 83 years for Swedish-speaking women. These numbers slightly exceed even those for Sweden (Statistics Finland, 2005b).

The table also shows that for women, most of the differential in life expectancy between the two population groups can be attributed to the situation at higher ages. For men, about one third each can be related to ages under 50 years, 50 to 69 years, and over 69 years, respectively.

Thus the mortality differential between Finnish speakers and Swedish speakers in Finland has not levelled off during the past decades. It is specifically pronounced in middle-aged men, but in higher ages it is about the same for men and women. This initial investigation also clearly reveals that, in order to understand why the two population groups differ in death rates, one should focus on higher ages.

3. DATA AND METHODS The data used for this part of our study come from the longitudinal

census data file compiled by Statistics Finland (Statistics Finland, 2005c). The version of the file utilised here consists of linked individual information for all Finnish residents from the censuses of 1970, 1975, 1980, 1985, 1990 and 1995. Data take the form of a multidimensional matrix that includes all individuals and information about their year of birth, birth region, current region of residence, sex, mother tongue, educational level, marital status and family situation.

Since the register is based on linked cross-sectional data it contains no explicit information about any events, and consequently not about causes of death. However, a person who disappears from the data must have died or emigrated. Since from ordinary statistics and supplementary data we know that migration has been a very rare event as compared with death in high ages, we can indirectly observe the event of death.

Considering this, the youngest age group that can be included in analysis consists of persons aged 60-64 years at the beginning of a five-year

Page 8: Regional mortality variation in Finland: a study of two

Tab

le 1

– L

ife e

xpec

tanc

y at

diff

eren

t age

s in

197

6 to

200

2 by

sex

and

pop

ulat

ion

grou

p

M

en

W

omen

Tim

e pe

riod

Tim

e pe

riod

76-8

2 83

-89

90-9

6 97

-02

76

-82

83-8

9 90

-96

97-0

2

F

inni

sh s

peak

ers

at a

ge

0

68

.5

70.3

71

.8

73.8

77.5

78

.5

79.5

80

.9

30

40.4

41

.9

43.4

45

.1

48

.7

49.6

50

.5

51.8

40

31

.3

32.8

34

.3

35.8

39.0

39

.9

40.9

42

.1

50

22.7

24

.1

25.6

27

.1

29

.7

30.5

31

.5

32.7

60

15

.4

16.5

17

.6

19.0

20.8

21

.6

22.5

23

.7

70

9.6

10

.4

11.0

11

.9

13

.0

13.6

14

.3

15.3

80

5

.2

6.0

6

.2

6.6

7.2

7

.5

7.8

8

.3

Swed

ish

spea

kers

at a

ge

0

71.4

73

.6

74.7

77

.0

78

.7

80.0

80

.7

82.2

30

42

.9

44.8

46

.0

48.0

49.5

50

.8

51.6

52

.8

40

33.5

35

.2

36.5

38

.4

39

.8

41.1

41

.9

43.1

50

24

.6

26.1

27

.4

29.2

30.5

31

.7

32.4

33

.5

60

16.8

17

.9

19.0

20

.6

21

.6

22.7

23

.3

24.5

70

10

.3

11.2

11

.9

13.0

13.7

14

.6

15.0

16

.0

80

5.6

6

.4

6.6

7

.2

7

.4

7.9

8

.3

9.0

D

iffer

ence

(S-F

) at a

ge

0

2

.9

3.3

2

.9

3.2

1.2

1

.5

1.2

1

.2

30

2.5

2

.8

2.6

2

.9

0

.8

1.2

1

.1

1.0

40

2

.3

2.5

2

.2

2.6

0.8

1

.2

1.0

1

.0

50

1.9

2

.0

1.8

2

.1

0

.8

1.1

0

.8

0.8

60

1

.4

1.4

1

.3

1.6

0.8

1

.1

0.8

0

.8

70

0.7

0

.8

0.8

1

.0

0

.6

0.9

0

.7

0.8

N

ote:

Cal

cula

tions

hav

e be

en b

ased

on

ordi

nary

vita

l sta

tistic

s.

176

Page 9: Regional mortality variation in Finland: a study of two

REGIONAL MORTALITY VARIATION IN FINLAND…

177

follow-up period. Published vital statistics indicate that in this age group the annual number of emigrants has been at most two to three per cent of the number of deaths. Restrictions of the data to even older people will therefore not change the results to be presented here, which is an argument confirmed also by our preliminary analysis. The highest age group studied is people aged 90-94 years.

Persons included in the analysis were consequently born between 1876 and 1930. The total number of observations is 3,976,137, representing 1,454,234 individuals (of which 1,324,384 are Finnish speakers and 129,850 are Swedish speakers). The total number of estimated deaths in each population group is 794,418 and 83,148.

Since death rates in ages under study are high, we estimate logistic regression models of five-year age-specific probabilities of dying, which include the variables “Age group”, “Year”, “Population group”, “Educational level”, “Family type”, “Birth region” and “Region of residence”. The model choice implies that observations pertaining to one individual are considered unrelated, and individuals are therefore not followed over time (as in a hazard model, for instance).

The unique mother tongue of each person is included in the population register in Finland. This information is used to distinguish the two population groups. Persons speaking foreign languages (and Sami), a total of 0.3 per cent, are excluded from our analysis.

Year expresses the time point and age group the five-year age-interval at the beginning of each follow-up period. Both these variables are used as categorical covariates.

Educational level comprises six categories that range from basic to graduate.

Family type has been obtained by combining information about marital status and family situation, and is used to reflect specific circumstances that prevail in older people. The variable consists of the categories “never married single”, “previously married single”, “living with spouse (and children)”, and “living with children, without spouse”. Single status means that a person is not living in a family (not with a spouse or with any children).

Birth region and region of residence are classified according to the former twelve regions of Finland (Figure 5). The Helsinki area, which consists of the cities Helsinki, Espoo, Vantaa and Kauniainen, has however been assigned a separate category, because mortality levels are markedly higher there than in the rest of Uusimaa. For birth region we additionally have a category that includes foreign-born. This will be referred to as “Ceded Karelia”, as the great majority of these people evidently were born in

Page 10: Regional mortality variation in Finland: a study of two

JAN SAARELA – FJALAR FINNÄS

178

that area, which belonged to Finland until 1944. Practically all Swedish speakers in Finland live concentrated at the southern and western coastline. For regions with Swedish-speaking settlements (Uusimaa, Turku & Pori, Vaasa and Åland) we therefore, for the purpose of specific analyses of the Swedish-speaking population, use a more detailed geographical categorisation (Figure 6).

Since the data constitute the whole population for a period of 25 years, significance tests in a traditional manner are inappropriate. Still, in order to indicate uncertainty due to population size, we occasionally comment on standard errors of the estimates, and in some instances also explicitly report risk populations.

Throughout the paper we report the parameters in exponentiated form, and therefore indicate only the size of standard errors. Confidence intervals for exponentiated parameters are generally not symmetric, but that is not a sensitive issue here, due to the small size of our standard errors1.

4. RESULTS OF STATISTICAL ANALYSIS 4.1 Mortality at the national level

We begin by observing the Finnish-Swedish mortality ratio at the

national level. The dependent variable is the odds of dying during a five-year follow-up period. For the sake of simplicity, however, we refer to it as “the death risk”.

Model 1 (in Table 2 for men and in Table 3 for women) controls only for age and observation year. In that case the Finnish-Swedish mortality differential is about 15 per cent for both men and women. When educational level and family type are added (Model 2 and Model 3), the ratio reduces marginally: for men with three percentage points and for women with only one percentage point. However, both these variables improve the model fit significantly and their estimated impact on mortality are very much as expected. For a person with graduate education, say, the death risk is about 30 per cent lower than for a person with basic education only. Likewise, a man who lives with a spouse has a 32 per cent lower death risk than a single man who has never been married. For women, the corresponding difference is 22 per cent.

1 In terms of confidence intervals; ( ) ( ))21(),21(, 22 σσσσ +−≈+− yyyy eeee , because

1−≈ xex for any small x.

Page 11: Regional mortality variation in Finland: a study of two

REGIONAL MORTALITY VARIATION IN FINLAND…

179

Figure 5 – Map of (former) regions of Finland

1

34

5

6

7

8

1011

12

13

14

9

1: Turku & Pori 2: Åland 3: Helsinki area 4: Uusimaa excluding Helsinki area 5: Vaasa 6: Häme 7: Kymi 8: Mikkeli 9: Central Finland10: Kuopio11: North Karelia12: Oulu13: Lapland14: Ceded Karelia

2

Page 12: Regional mortality variation in Finland: a study of two

JAN SAARELA – FJALAR FINNÄS

180

Figure 6 – Map of the main settlements area of Swedish speakers in Finland (the coloured parts)

12

34

5

6

7

8

1: Turunmaa2 Åland:3: Helsinki area4: Western Uusimaa5: Eastern Uusimaa6: Southern Pohjanmaa7: Central Pohjanmaa8: Northern Pohjanmaa9: All other areas

9

To study regional mortality differentials, birth region and region of

residence are included one at a time (Model 4 and Model 5), and then both simultaneously (Model 6). They display the known pattern with gradually increasing death rates in the South-West to North-East direction. The Helsinki area although deviates in this respect. Specifically men in Helsinki have remarkably high death rates. Birth region is evidently more important than region of residence, according to the reduction in the log likelihood. Variation across estimates is also larger for birth region, but the Finnish-Swedish mortality ratio further decreases when region of residence is

Page 13: Regional mortality variation in Finland: a study of two

Tab

le 2

– M

orta

lity

ratio

mod

els

for

men

in F

inla

nd

M

odel

1

Mod

el 2

M

odel

3

Mod

el 4

M

odel

5

Mod

el 6

P

opul

atio

n gr

oup

Fin

nish

spe

aker

(ref

.)

1

1

1

1

1

1

S

wed

ish

spea

ker

0.8

44

0.8

74

0.8

78

0.8

96

0.8

90

0.8

98

A

ge g

roup

6

0-64

yea

rs (r

ef.)

1

1

1

1

1

1

65-

69

1.5

98

1.5

90

1.5

77

1.5

78

1.5

78

1.5

79

70-

74

2.5

90

2.5

68

2.5

01

2.5

05

2.5

06

2.5

08

75-

79

4.4

25

4.3

73

4.1

23

4.1

34

4.1

35

4.1

41

80-

84

7.8

51

7.7

51

6.9

84

7.0

08

7.0

15

7.0

29

85-

89

14.4

94

14.3

10

12.2

14

12.2

73

12.2

91

12.3

20

90-

94

32.3

30

31.9

25

25.8

62

25.9

99

26.0

35

26.1

08

Ye

ar

197

0 (r

ef.)

1

1

1

1

1

1

197

5 0

.866

0

.871

0

.873

0

.873

0

.873

0

.873

1

980

0.8

04

0.8

15

0.8

14

0.8

15

0.8

14

0.8

14

198

5 0

.757

0

.772

0

.779

0

.779

0

.778

0

.778

1

990

0.6

63

0.6

81

0.6

85

0.6

86

0.6

85

0.6

85

E

duca

tiona

l lev

el

Bas

ic (r

ef.)

1

1

1

1

1

Low

er le

vel l

ower

voc

atio

nal

0

.796

0

.819

0

.817

0

.819

0

.816

U

pper

leve

l low

er v

ocat

iona

l

0.7

87

0.8

16

0.8

04

0.8

12

0.8

03

Upp

er v

ocat

iona

l

0.6

79

0.7

11

0.7

04

0.7

10

0.7

04

Und

ergr

adua

te

0

.750

0

.785

0

.763

0

.779

0

.762

G

radu

ate

0

.663

0

.701

0

.677

0

.694

0

.677

181

Page 14: Regional mortality variation in Finland: a study of two

Tab

le 2

– c

ont’d

Mod

el 1

M

odel

2

Mod

el 3

M

odel

4

Mod

el 5

M

odel

6

Fam

ily ty

pe

Nev

er m

arrie

d si

ngle

(ref

.)

1

1

1

1

P

revi

ousl

y m

arri

ed s

ingl

e

1

.028

1

.025

1

.028

1

.025

L

ivin

g w

ith s

pous

e (a

nd c

hild

ren)

0

.672

0

.672

0

.672

0

.672

L

ivin

g w

ith c

hild

ren,

with

out s

pous

e

0

.770

0

.771

0

.770

0

.771

Reg

ion

of r

esid

ence

T

urku

& P

ori (

ref.)

1

1

Å

land

0.9

98

1

.104

H

elsi

nki a

rea

1

.147

1.1

00

Uus

imaa

exc

ludi

ng H

elsi

nki a

rea

1

.053

1.0

30

Vas

a

0.9

85

0

.963

H

äme

1

.050

1.0

28

Kym

i

1.0

86

1

.033

M

ikke

li

1.0

92

1

.036

C

entr

al F

inla

nd

1

.079

1.0

25

Kuo

pio

1

.104

1.0

11

Nor

th K

arel

ia

1

.140

1.0

54

Oul

u

1.0

68

1

.004

L

apla

nd

1

.070

1.0

26

182

Page 15: Regional mortality variation in Finland: a study of two

Tab

le 2

– c

ont’d

Mod

el 1

M

odel

2

Mod

el 3

M

odel

4

Mod

el 5

M

odel

6

Bir

th r

egio

n

T

urku

& P

ori (

ref.)

1

1

Å

land

0

.990

0

.906

H

elsi

nki a

rea

1.1

85

1.1

11

Uus

imaa

exc

ludi

ng H

elsi

nki a

rea

1.0

62

1.0

24

Vaa

sa

1.0

12

1.0

43

Häm

e

1

.050

1

.027

K

ymi

1.1

06

1.0

77

Mik

keli

1.1

15

1.0

85

Cen

tral

Fin

land

1

.105

1

.084

K

uopi

o

1

.159

1

.147

N

orth

Kar

elia

1

.170

1

.125

O

ulu

1.1

00

1.1

00

Lap

land

1

.081

1

.063

C

eded

Kar

elia

1

.116

1

.098

Con

stan

t 0

.209

0

.217

0

.292

0

.275

0

.269

0

.266

Log

like

lihoo

d -8

03,5

72

-802

,074

-7

97,6

13

-797

,321

-7

97,2

88

-797

,190

Not

e:

The

sta

ndar

d er

rors

are

roug

hly

one

per c

ent o

f the

est

imat

es, i

.e. a

bout

0.0

1 fo

r a p

aram

eter

of s

ize

1 an

d ab

out 0

.2 fo

r a

para

met

er o

f siz

e 20

.

183

Page 16: Regional mortality variation in Finland: a study of two

Tab

le 3

– M

orta

lity

ratio

mod

els

for

wom

en in

Fin

land

Mod

el 1

M

odel

2

Mod

el 3

M

odel

4

Mod

el 5

M

odel

6

Pop

ulat

ion

grou

p

F

inni

sh s

peak

er (r

ef.)

1

1

1

1

1

1

Sw

edis

h sp

eake

r 0

.852

0

.864

0

.865

0

.895

0

.892

0

.898

Age

gro

up

60-

64 y

ears

(ref

.)

1

1

1

1

1

1

6

5-69

1

.786

1

.777

1

.728

1

.729

1

.730

1

.730

7

0-74

3

.396

3

.364

3

.184

3

.187

3

.191

3

.191

7

5-79

6

.703

6

.622

6

.111

6

.122

6

.134

6

.135

8

0-84

13

.484

13

.298

12

.027

12

.064

12

.097

12

.102

8

5-89

27

.538

27

.142

24

.199

24

.301

24

.401

24

.413

9

0-94

59

.419

58

.509

51

.748

52

.032

52

.303

52

.344

Year

1

970

(ref

.)

1

1

1

1

1

1

1

975

0.7

65

0.7

68

0.7

71

0.7

71

0.7

70

0.7

70

198

0 0

.698

0

.703

0

.707

0

.707

0

.706

0

.706

1

985

0.6

79

0.6

88

0.6

96

0.6

96

0.6

95

0.6

95

199

0 0

.599

0

.610

0

.620

0

.620

0

.619

0

.618

Edu

catio

nal l

evel

B

asic

(ref

.)

1

1

1

1

1

L

ower

leve

l low

er v

ocat

iona

l

0.8

11

0.8

09

0.8

08

0.8

07

0.8

07

Upp

er le

vel l

ower

voc

atio

nal

0

.778

0

.768

0

.769

0

.767

0

.770

U

pper

voc

atio

nal

0

.733

0

.727

0

.726

0

.727

0

.727

U

nder

grad

uate

0.7

13

0.7

00

0.7

03

0.7

00

0.7

03

Gra

duat

e

0.6

80

0.6

66

0.6

69

0.6

67

0.6

70

184

Page 17: Regional mortality variation in Finland: a study of two

Tab

le 3

– c

ont’d

Mod

el 1

M

odel

2

Mod

el 3

M

odel

4

Mod

el 5

M

odel

6

Fam

ily ty

pe

Nev

er m

arri

ed s

ingl

e (r

ef.)

1

1

1

1

Pre

viou

sly

mar

ried

sin

gle

0.9

78

0.9

72

0.9

73

0.9

71

Liv

ing

with

spo

use

(and

chi

ldre

n)

0.7

79

0.7

75

0.7

77

0.7

75

Liv

ing

with

chi

ldre

n, w

ithou

t spo

use

0.8

69

0.8

59

0.8

62

0.8

58

R

egio

n of

res

iden

ce

Tur

ku &

Por

i (re

f.)

1

1

Åla

nd

0

.881

0.9

62

Hel

sink

i are

a

1.0

47

1

.014

U

usim

aa e

xclu

ding

Hel

sink

i are

a

1.0

58

1

.043

V

asa

1

.014

1.0

05

Häm

e

1.0

31

1

.021

K

ymi

1

.118

1.0

64

Mik

keli

1

.139

1.0

82

Cen

tral

Fin

land

1.1

22

1

.078

K

uopi

o

1.1

09

1

.038

N

orth

Kar

elia

1.1

68

1

.086

O

ulu

1

.081

1.0

34

Lap

land

1.1

13

1

.084

185

Page 18: Regional mortality variation in Finland: a study of two

Tab

le 3

– c

ont’d

Mod

el 1

M

odel

2

Mod

el 3

M

odel

4

Mod

el 5

M

odel

6

Bir

th r

egio

n

T

urku

& P

ori (

ref.)

1

1

Å

land

0

.880

0

.906

H

elsi

nki a

rea

1.1

05

1.0

87

Uus

imaa

exc

ludi

ng H

elsi

nki a

rea

1.0

34

1.0

05

Vaa

sa

1.0

27

1.0

21

Häm

e

1

.018

1

.002

K

ymi

1.1

31

1.0

82

Mik

keli

1.1

41

1.0

83

Cen

tral

Fin

land

1

.122

1

.068

K

uopi

o

1

.145

1

.110

N

orth

Kar

elia

1

.183

1

.119

O

ulu

1.1

06

1.0

75

Lap

land

1

.106

1

.037

C

eded

Kar

elia

1

.117

1

.089

Con

stan

t 0

.090

0

.093

0

.108

0

.102

0

.099

0

.099

Log

like

lihoo

d -1

,030

,449

-1

,029

,317

-1

,027

,829

-1

,027

,487

-1

,027

,387

-1

,027

,315

Not

e:

The

sta

ndar

d er

rors

are

roug

hly

one

per c

ent o

f the

est

imat

es, i

.e. a

bout

0.0

1 fo

r a p

aram

eter

of s

ize

1 an

d ab

out 0

.2 fo

r a

para

met

er o

f siz

e 20

.

186

Page 19: Regional mortality variation in Finland: a study of two

REGIONAL MORTALITY VARIATION IN FINLAND…

187

accounted for. There is an additional improvement of the model fit when both variables are included, but the additional reduction in the Finnish-Swedish mortality ratio is only a couple of percentage points. Relative differences between western regions and eastern regions are about the same as between Finnish speakers and Swedish speakers.

In Section 2 of this paper, we saw that the difference in men is lower at older ages than at younger ones. This is confirmed by these data for ages over 60. The interaction between population group and age also significantly improves the model fit, but it does not change the estimated effects for the other variables. Throughout the analysis we have tested for interaction effects, but in order to provide a brief overview we present and discuss them explicitly only on occasions of specific interest.

The findings so far give rise to two additional main questions, which need to be studied separately. One is whether the Finnish-Swedish mortality differential varies across geographical regions. The other concerns the interrelation between (internal) migration and mortality. These issues are important due to the fact that Swedish speakers live in distinct areas along the coast, and because the internal migration behaviour of the two population groups have differed substantially. Thus internal migration in Finland has caused substantial geographical redistribution of the cohorts studied, and specifically of the Finnish speakers since they have had much higher internal migration rates than the Swedish speakers. A consequence is that a higher proportion of Swedish speakers compared to Finnish speakers live in the region in which they were born, and that regions with both Swedish-speaking and Finnish-speaking inhabitants differ greatly with regard to in-migration.

To see if the between-group mortality differential varies across spatial areas, we estimate separate regressions for each region with both Swedish-speaking and Finnish-speaking settlements. The interrelation between migration and mortality is studied by focusing on the Finnish speakers and thus including the whole country into analysis.

4.2 Geographical variation in the Finnish-Swedish mortality differential

The results for models in which we have estimated separate regressions

for the four regions with both Swedish-speaking and Finnish-speaking inhabitants are summarised in Table 4. It gives the population-group mortality ratio in models with various sets of control variables.

We can see that the Finnish-Swedish mortality differential varies regionally. It is fairly small in Turku & Pori and in Uusimaa, but substantially larger in the Helsinki area and in Vaasa. In the first two

Page 20: Regional mortality variation in Finland: a study of two

JAN SAARELA – FJALAR FINNÄS

188

Table 4 – Swedish-Finnish mortality ratio by region of residence in alternative models

Adjusted for Turku & Pori

Helsinki area

Uusi- maa Vaasa

Men

Age,Year 0.931 0.856 0.895 0.833 + Educational level 0.973 0.889 0.906 0.845 + Family type 0.965 0.900 0.906 0.855 + Birth region 0.987 0.899 0.942 0.869 + Data restriction: born in present region of residence 0.982 0.878 0.957 0.866

Women Age,Year 0.912 0.910 0.884 0.861

+ Educational level 0.930 0.923 0.890 0.866 + Family type 0.932 0.929 0.892 0.868 + Birth region 0.944 0.931 0.937 0.880 + Data restriction: born in present region of residence 0.962 0.917 0.940 0.877

Note: Standard errors of the unexponentiated estimates are at most 0.025

regions, the death risk of a male Swedish speaker is less than five per cent lower than that of a male Finnish speaker, whereas in the latter two roughly 12 per cent lower. For women there is a similar pattern, albeit the numbers are somewhat different in magnitude. As before, the Finnish-Swedish mortality differential reduces somewhat when educational level and family type are accounted for. The contribution of birth region to the difference in death risks between population groups varies across regions, being large in Turku & Pori and in Uusimaa, much smaller in Vaasa, and practically nil in the Helsinki area. To further refine the comparison, and in an attempt to avoid potential effects of selective migration, we also try to restrict the data to persons born in the present region of residence (the final row). This course of action does not change the results to any noteworthy degree, however.

Regional mortality differentials within the Swedish-speaking population group cannot be seen directly from Table 4. Therefore we also perform separate regressions for Swedish speakers only, using the more detailed geographical division referred to in Section 3. Results of these regressions are summarised in Table 5 for men and women separately. They tell us that there tends to be geographical variation in death rates also within the Swedish-speaking population. The death risk appears to be lowest in Åland and gradually higher in the eastern and northern directions. However, the

Page 21: Regional mortality variation in Finland: a study of two

REGIONAL MORTALITY VARIATION IN FINLAND…

189

pattern is not monotonous and mortality is markedly higher in the Helsinki area than elsewhere. Both region of residence and birth region improve the model fit, and the impact of control variables is very much as for the whole population.

4.3 Birth region, region of residence and mortality

The question of whether there is any interrelation between internal

migration and mortality is closely related to the issue of whether one can observe a “healthy migrant effect” (Marmot et al., 1984; Friis et al., 1998). If that is the case within-country migrants would be inherently healthier and thus have lower mortality rates than non-migrants. The literature is largely silent, however, on whether better health due to selection reflects superior health habits, behaviours and conditions in the sending areas (countries), or whether it is principally because of health selectivity among migrants compared to those who stayed (Jasso et al., 2004). However, Marmot et al. (1984) suggest that the health selection effect may be stronger the higher the costs of moving are. As the costs of moving within a country are relatively low, at least compared with costs associated with international migration, it is plausible that we cannot observe a “healthy migrant effect”. In our opinion, one could equally well argue that within-country migrants are less integrated in the local society than the “local natives”. Such social disintegration may lead to feelings of alienation and social isolation, which can affect well-being and health (Kawachi and Kennedy, 1997; Kawachi et al., 1997), and thus, in the long-run, mortality. This suggests that the within-country migrants might have high relative mortality levels.

We accomplish this part of our analysis by interacting birth region with region of residence. The results of models which account for such interaction effects, and control for the impact of background variables, are summarised by sex and population group in Tables A1 to A4 in the Appendix. In spite of the fact that the number of observations is small (see Tables A5 to A8 in the Appendix), clear patterns can be seen. However, to facilitate interpretation we combine all persons who have out-migrated from a region into one category, and all persons who in-migrated into a region into one category. Thus for each region we distinguish “natives”, “out-migrants” and “in-migrants”. It should be borne in mind, however, that the time of migration is not known from the data.

The results of the estimations are summarised and presented in the maps in Figure 7 to Figure 10. For each region (or unit), the first number gives relative mortality of people born and residing in the region. The second is for people born in the region who have out-migrated to any other region, and the

Page 22: Regional mortality variation in Finland: a study of two

Tab

le 5

– M

orta

lity

ratio

mod

els

for

Swed

ish

spea

kers

in F

inla

nd

Men

Wom

en

M

odel

1

Mod

el 2

M

odel

3

M

odel

1

Mod

el 2

M

odel

3

Age

gro

up

6

0-64

yea

rs (r

ef.)

1

1

1

1

1

1

6

5-69

1

.615

1

.619

1

.617

1.5

84

1.5

86

1.5

85

70-

74

2.6

43

2.6

54

2.6

48

2

.882

2

.888

2

.884

7

5-79

4

.492

4

.518

4

.505

5.5

15

5.5

33

5.5

22

80-

84

7.9

58

8.0

18

7.9

95

10

.842

10

.892

10

.865

8

5-89

14

.453

14

.581

14

.532

22.4

93

22.6

24

22.5

59

90-

94

31.8

29

32.1

77

32.0

27

47

.504

47

.871

47

.697

Ye

ar

1

970

(ref

.)

1

1

1

1

1

1

197

5 0

.820

0

.819

0

.820

0.7

79

0.7

78

0.7

78

198

0 0

.794

0

.790

0

.793

0.7

52

0.7

50

0.7

52

198

5 0

.754

0

.749

0

.752

0.7

42

0.7

39

0.7

41

199

0 0

.673

0

.667

0

.671

0.6

71

0.6

67

0.6

70

Edu

catio

nal l

evel

Bas

ic (r

ef.)

1

1

1

1

1

1 L

ower

leve

l low

er v

ocat

iona

l 0

.789

0

.793

0

.789

0.8

58

0.8

57

0.8

55

Upp

er le

vel l

ower

voc

atio

nal

0.8

63

0.8

79

0.8

62

0

.789

0

.793

0

.786

U

pper

voc

atio

nal

0.7

44

0.7

53

0.7

42

0

.800

0

.801

0

.799

U

nder

grad

uate

0

.775

0

.797

0

.775

0.7

30

0.7

35

0.7

26

Gra

duat

e 0

.682

0

.699

0

.678

0.6

96

0.7

01

0.6

93

Fam

ily ty

pe

N

ever

mar

ried

sing

le

1

1

1

1

1

1 P

revi

ousl

y m

arri

ed s

ingl

e 1

.020

1

.024

1

.018

0.9

33

0.9

33

0.9

33

Liv

ing

with

spo

use

(and

chi

ldre

n)

0.6

88

0.6

90

0.6

88

0

.747

0

.746

0

.747

L

ivin

g w

ith c

hild

ren,

with

out s

pous

e 0

.850

0

.848

0

.849

0.8

61

0.8

59

0.8

61

190

Page 23: Regional mortality variation in Finland: a study of two

Tab

le 5

– c

ont’d

M

en

W

omen

Mod

el 1

M

odel

2

Mod

el 3

Mod

el 1

M

odel

2

Mod

el 3

R

egio

n of

res

iden

ce

T

urun

maa

1

1

1

1

Åla

nd

0.9

17

1

.042

0.8

34

0

.928

H

elsi

nki a

rea

1.0

57

1

.022

1.0

17

1

.036

W

este

rnU

usim

aa

0.9

57

0

.955

1.0

06

1

.045

E

aste

rn U

usim

aa

1.0

30

0

.989

1.0

05

1

.000

S

outh

ern

Pohj

anm

aa

0.8

26

0

.843

0.9

62

1

.043

C

entr

al P

ohja

nmaa

0

.924

0.8

97

0

.954

0.9

45

Nor

ther

n Po

hjan

maa

0

.907

0.8

85

0

.945

0.9

38

All

othe

r are

as

0.9

68

0

.947

0.9

56

0

.970

B

irth

reg

ion

T

urun

maa

1

1

1

1

Åla

nd

0

.893

0

.858

0

.810

0

.868

H

elsi

nki a

rea

1

.082

1

.073

1

.004

0

.981

W

este

rnU

usim

aa

0

.963

0

.978

0

.948

0

.921

E

aste

rn U

usim

aa

1

.046

1

.049

1

.008

1

.033

S

outh

ern

Pohj

anm

aa

0

.823

0

.955

0

.882

0

.862

C

entr

al P

ohja

nmaa

0.9

58

1.0

51

0.9

89

1.0

39

Nor

ther

n Po

hjan

maa

0.9

39

1.0

40

0.9

82

1.0

40

All

othe

r are

as

0

.951

1

.018

0

.965

0

.971

C

onst

ant

0.2

53

0.2

50

0.2

52

0

.102

0

.103

0

.103

L

og li

kelih

ood

-73,

657

-73,

665

-73,

644

-9

8,14

1 -9

8,13

7 -9

8,12

0

N

ote:

T

he s

tand

ard

erro

rs a

re ro

ughl

y th

ree

per c

ent o

f the

est

imat

es, i

.e. a

bout

0.0

3 fo

r a p

aram

eter

of s

ize

1 an

d ab

out 0

.6 fo

r a

para

met

er o

f siz

e 20

.

191

Page 24: Regional mortality variation in Finland: a study of two

JAN SAARELA – FJALAR FINNÄS

192

third is for people who have in-migrated to that region. For Finnish speakers, people born and residing in the region of Turku & Pori constitute the reference category (100), whereas people born and residing in Turunmaa constitute the index group for Swedish speakers.

For native Finnish speakers there is a fairly evident pattern in support of an accentuation of the death risk in the South-West to North-East direction. An exception is natives in the Helsinki area. As compared with natives in Turku & Pori, for instance, they have 24 per cent higher death risks. For out-migrants there is a similar geographical variation in death rates as for natives. People born in regions with relatively high mortality levels consequently tend to take these high death risks also if moving elsewhere. For in-migrants, who constitute a mixture of geographical origin, there is no systematic geographical pattern in death rates.

In men, mortality levels of out-migrants and in-migrants are generally higher than those of natives. Out-migrants from the Helsinki area although have lower death rates than native Helsinki citizens. Still, there appears to be gender differentials. Women who have in-migrated to the Helsinki area, for instance, have relatively low mortality levels.

As internal migration has been much lower in the Swedish-speaking population than in the Finnish-speaking one, the number of observations is smaller, and the results for Swedish speakers thus less systematic and more difficult to interpret. Before 1970, the proportion of people who emigrated abroad in any given birth cohort was substantially higher in Swedish speakers than in Finnish speakers, and particularly high in certain parts of the Swedish speakers’ settlement area such as Pohjanmaa and Åland (Finnäs, 1986). For people categorised as out-migrants in our context, this may result in somewhat ambiguous parameters with regard to regional mortality variation.

The highest mortality rate is found for people born and residing in the Helsinki area, whereas it is markedly lower for those who have out-migrated from Helsinki. People originating from Åland and Pohjanmaa have the lowest mortality rates, whereas those from the eastern parts of the settlement area (Eastern Uusimaa) have relatively high mortality levels. There is consequently a geographical pattern that to a certain, but fairly limited, extent is similar to that of the Finnish speakers.

Page 25: Regional mortality variation in Finland: a study of two

REGIONAL MORTALITY VARIATION IN FINLAND…

193

Figure 7 – Relative mortality by region, Finnish-speaking men

108123115

111118114

116123114

118122121

102109116 110

118116

113116117

–115–

111116115

106110113100

106110

124112117

103109113

Born and residing in the regionOut-migrated from the regionIn-migrated to the region

Note: born and residing in Turku & Pori=100.

Page 26: Regional mortality variation in Finland: a study of two

JAN SAARELA – FJALAR FINNÄS

194

Figure 8 – Relative mortality by region, Finnish-speaking women

112109117

112111111

115116114

120118121

104102112 116

109116

118111116

–113–

116111115

103101109100

100107

112111106

103101112

Born and residing in the regionOut-migrated from the regionIn-migrated to the region

Note: born and residing in Turku & Pori=100.

Page 27: Regional mortality variation in Finland: a study of two

REGIONAL MORTALITY VARIATION IN FINLAND…

195

Figure 9 – Relative mortality by area, Swedish-speaking men

100104108 96

100101

111104104

106107103

9089

109

84104111

959893

939692

Born and residing in the regionOut-migrated from the regionIn-migrated to the region

100101

97

Note: born and residing in Turunmaa=100.

Page 28: Regional mortality variation in Finland: a study of two

JAN SAARELA – FJALAR FINNÄS

196

Figure 10 – Relative mortality by area, Swedish-speaking women

100103

99 9794

108

10394

100

102102

9680

10095

9582

117

96107

93

969593

Born and residing in the regionOut-migrated from the regionIn-migrated to the region

97100

94

Note: born and residing in Turunmaa=100.

Page 29: Regional mortality variation in Finland: a study of two

REGIONAL MORTALITY VARIATION IN FINLAND…

197

5. SUMMARY AND CONCLUSIONS The primary focus of this paper has been to explore the Finnish-

Swedish mortality differential in terms of accounting for regional mortality variation and specifically a person’s region of birth. Thus the paper has been concerned with two of the perhaps most distinctive features of overall mortality in Finland, namely the variation in death rates between population groups and across regions.

Using vital statistics, we show that the relative difference in death risks between Finnish speakers and Swedish speakers has remained fairly constant during the past 30 years. High ages account for a considerable part of the between-group mortality differential.

Census register data that cover a 25-year period are further used to analyse how mortality relates to birth region and region of residence in people aged 60 years and over. In line with the results of previous studies, we find that death rates in the Finnish-speaking population increase gradually in the South-West to North-East direction. Even more detailed analyses further reveal that there is a somewhat similar regional mortality variation also within the Swedish-speaking population, albeit much less emphasised in pattern. Death rates in the Helsinki area are still found to be remarkably high. These findings echo previous results and could be caused by a number of complex factors that affect health in larger cities, such as socio-economic distances, population density, pollution and violence (Valkonen and Kauppinen, 2001; Galea et al., 2005).

Since Swedish speakers in Finland live geographically concentrated in parts of the country where overall death rates are relatively low, mortality differences in relation to Finnish speakers, as observed at the national level, are partly interrelated with region of residence. Still, even after restricting the analyses to regions with both Swedish-speaking and Finnish-speaking settlements a mortality differential remains. Controlling for birth region and a number of other covariates additionally helps to reduce the Finnish-Swedish mortality ratio in the statistical models, but in some areas a large difference remains.

Our results indicate that some other individual-related characteristics, besides those adjusted for here, are important for mortality. Having run regressions that account for the interaction between birth region and region of residence, we find that the geographical mortality pattern based on out-migrants’ birth region resembles that of people who have not migrated. This indicates that circumstances operating before or in childhood can be important. Additional support for the view that factors other than present environmental conditions impact on mortality is that there is no systematic

Page 30: Regional mortality variation in Finland: a study of two

JAN SAARELA – FJALAR FINNÄS

198

geographical pattern in death rates of in-migrants. Also in a strict statistical sense, birth region is found to be a generally more decisive determinant of mortality than region of residence.

We cannot find any indications of a “healthy migrant effect”. On the contrary, our results clearly reveal that within-country migrants have higher relative mortality levels than non-migrants. It is far from obvious what factors govern these results, but since we account for geographical founders, it is plausible that people who do not have their geographical roots in the present region of residence are less integrated in the local society. That could affect health and thus mortality.

In light of the substantial improvement in living standards and public health during the 20th century, the significant role of birth region on mortality must be considered a very striking issue. As an indicator of people’s geographical extraction, the variable may still be considered fairly crude. Mortality variation by birth region may therefore reflect an array of factors, including life-time environmental conditions, nutritional factors, childhood socio-economic circumstances, access to health care and levels of social cohesion. The current economic situation does not provide any explanation, however. We originally tried to use a supportive sample that includes detailed information about individuals’ income. On the basis of those data we found no indications whatsoever that income differences would underlie the Finnish-Swedish mortality differential, in spite of the fact that overall mortality decreased notably with income.

Circumstances that prevail at childhood, such as socio-economic conditions, diet intake, health behaviour, or other environmental factors, may gradually lose importance when a person migrates into a new environment. If that is the case, the question is why birth region has such strong an impact on mortality. Considering the nature of our data, no definite answer can be given.

We still want to point out that official public statistics reveal that variation in infant mortality in the beginning of the 1920s, i.e. during the time period in which part of the people in the data were born or grew up, has a fairly similar regional pattern to that observed here. In eastern Finland infant mortality was about 20 per cent higher than in western Finland, and specifically in Åland relative mortality was low. Our results might therefore be interpreted as reflecting variation in living standards and socio-economic conditions during the time the people under study were born.

It cannot be ruled out, however, that part of the mortality variation observed here is interrelated with differentials in recessive inheritance within the population of Finland. Thus similarities in regional mortality variation between out-migrants and natives may reflect the fact that an individual’s

Page 31: Regional mortality variation in Finland: a study of two

REGIONAL MORTALITY VARIATION IN FINLAND…

199

gene structure is time consistent. The genes of the population in Finland have been studied since the 1920s. Early studies found logical but small differences in the frequencies of ABO blood groups between eastern and western populations (Streng, 1935; 1937; Mustakallio, 1937). Later results suggested that the population of Finland can be divided into residents of Åland, West Finland and East Finland (Kajanoja, 1971; 1973). This would resemble the view of how the country was inhabited and populated according to archaeological and historical data (Workman et al., 1976; Norio, 2003). Studies have also shown that Swedish speakers in Finland tend to be genetically intermediate between people in Sweden and Finnish speakers in Finland (Kajanoja, 1972; Eriksson, 1973; Sahi, 1974; Eriksson et al., 1986), as well as that they differ genetically from the Finnish speakers (Virtaranta-Knowles et al., 1991). The degree of genetic admixture is claimed to be as high as 60 per cent and fairly homogenous, as it occurred at an early stage and language thereafter worked as a barrier to prevent it any further.

The present findings are not the first in demographic literature that point towards a potential link between ancestry and mortality. For instance, the elevated mortality of elderly people in Eastern Finland have been found to be totally attributed to a higher prevalence of coronary diseases (Koskinen and Martelin, 1998), whereas such deaths are substantially less common in areas dominated by Swedish-speaking settlements (Valkonen and Martikainen, 1990). In the event regional mortality variation is interrelated with genetic differences in the population of Finland, it is plausible that the residual variation between Finnish speakers and Swedish speakers observed here also have to do with similar circumstances. Thus there seems to be great opportunities for interdisciplinary research to broaden our understanding of mortality inequality in Finland.

Acknowledgements We are grateful for comments from anonymous referees, and seminar

participants at SOFI (Stockholm University), CeFAM (Karolinska Institutet), and the First Nordic Meeting on Register-based Health Research in Helsinki.

Page 32: Regional mortality variation in Finland: a study of two

JAN SAARELA – FJALAR FINNÄS

200

References

BLOMGREN J., MARTIKAINEN P., MÄKELÄ P., VALKONEN T. (2004), “The Effects of Regional Characteristics on Alcohol-Related Mortality – A Register-Based Multilevel Analysis of 1.1 Million Men”, Social Science & Medicine, 58, 2523-2535.

DOBLHAMMER G. (2004), The Late Life Legacy of Very Early Life, Berlin, Springer-Verlag.

ERIKSSON A.W. (1973), “Genetic Polymorphisms in Finno-Ugrian Populations”, Israel Journal of Medical Sciences, 9, 1156-1170.

ERIKSSON A.W., PARTANEN K., FRANTS R.R., PRONK J.C., KOSTENSE P.J. (1986), “ABH Secretion Polymorphism in Icelanders, Åland Islanders, Finns, Finnish Lapps, Komi and Greenland Eskimos: A Review and New Data”, Annals of Human Biology, 13, 273-285.

FINNÄS F. (1986), Den finlandssvenska befolkningsutvecklingen 1950-1980. En analys av en språkgrupps demografiska utveckling och effekten av blandäktenskap, Doctoral Dissertation, Skrifter utgivna av svenska litteratursällskapet i Finland, No. 533. Helsinki, Svenska litteratursällskapet i Finland.

FRIIS R., YNGVE A., PERSSON V. (1998), “Review of Social Epidemiologic Research on Migrants’ Health: Findings, Methodological Cautions, and Theoretical Perspectives”, Scandinavian Journal of Social Medicine, 26, 173-180.

GALEA S., FREUDENBERG N., VLAHOV D. (2005), “Cities and Population Health”, Social Science & Medicine, 60, 1017-1033.

Jasso G., Massey D.S., Rosenzweig M.R., Smith J.P. (2004), Immigrant Health: Selectivity and Acculturation, Working Paper Series, No. 23. London, The Institute for Fiscal Studies.

KAJANOJA P. (1971), “A Study in the Morphology of the Finns and its Relation to the Settlement of Finland”, Annales Academiae Scientiarum Fennicae, Series A, V Medica, 146.

KAJANOJA P. (1972), “A Contribution to the Physical Anthropology of the Finns”, Annales Academiae Scientiarum Fennicae, Series A, V Medica, 153.

KAJANOJA P. (1973), “Comparison of the Results from Anthropological Research on Finns”, in Studies in the Anthropology of the Finno-Ugrian Peoples, Papers, No. 7, 89-107. University of Helsinki, Department of Archaeology.

KAWACHI I., KENNEDY B.P. (1997), “Socioeconomic Determinants of Health: Health and Social Cohesion: Why Care About Income Inequality?”, British Medical Journal, 314, 1037.

Page 33: Regional mortality variation in Finland: a study of two

REGIONAL MORTALITY VARIATION IN FINLAND…

201

KAWACHI I., KENNEDY B.P., LOCHNER K., PROTHROW-STITH D. (1997), “Social Capital, Income Inequality, and Mortality”, American Journal of Public Health, 87, 1491-1498.

KOSKINEN S. (1994), Origin of Regional Differences in Mortality From Ischaemic Heart Disease in Finland, Doctoral Dissertation. Research Reports, No. 41. Jyväskylä, National Research and Development Centre for Welfare and Health.

KOSKINEN S., MARTELIN T. (1994), “Kuolleisuus”, in KOSKINEN S., MARTELIN T., NOTKOLA I.-L., NOTKOLA V., PITKÄNEN K. (eds.), Suomen Väestö, Hämeenlinna, Gaudeamus, 150-225.

KOSKINEN S., MARTELIN T. (1998), “Viina tappaa keski-ikäisiä pääkaupunkiseudulla. Kuolleisuuden alue-erot 1990-luvulla”, Kuntapuntari, 2, 66-71.

KOSKINEN S., MARTELIN T. (2003), “Why is Mortality Low among the Swedish-speaking Minority in Finland?”, Yearbook of Population Research in Finland, 39, 15-32.

LANGFORD I.H., BENTHAM G. (1996), “Regional Variations in Mortality Rates in England and Wales: An Analysis Using Multi-level Modeling”, Social Science & Medicine, 42, 897-908.

LUY M. (2004), “Mortality Differences between Western and Eastern Germany before and after Reunification. A Macro and Micro Level Analysis of Developments and Responsible Factors”, Genus, LX(3-4), 99-141.

MARMOT M.G., ADELSTEIN A.M., BULUSU L. (1984), “Lessons from the Study of Immigrant Mortality”, Lancet, 323, 1455-1457.

MARTELIN T. (1994), Differential Mortality at Older Ages, Doctoral Dissertation. Publications of the Finnish Demographic Society, No. 16. Helsinki.

MUSTAKALLIO E. (1937), “Untersuchungen über die M- N-, A1- A2- und O- A- B-Blutgruppen in Finnland”, Acta Societatis Medicorum Fennicae Duodecim, 20 A (20/4), 1-185.

NORIO R. (2003), “Finnish Disease Heritage II: Population Prehistory and Genetic Roots of Finns”, Human Genetics, 112, 457-469.

PITKÄNEN K., KOSKINEN S., MARTELIN T. (2000), “Kuolleisuuden alue-erot ja niiden historia”, Duodecim, 116, 1697-1710.

REGIDOR E., GUTIERREZ-FISAC J.L., CALLE M.E., NAVARRO P., DOMÍNGUEZ V. (2002), “Infant Mortality at Time of Birth and Cause-Specific Adult Mortality among Residents of the Region of Madrid Born Elsewhere in Spain”, International Journal of Epidemiology, 31, 368-374.

SAARELA J., FINNÄS F. (2005a), “Mortality Inequality in Two Native Population Groups”, Population Studies, 59, 313-320.

Page 34: Regional mortality variation in Finland: a study of two

JAN SAARELA – FJALAR FINNÄS

202

SAARELA J., FINNÄS F. (2005b), “Geographical Extraction and the Finnish-Swedish Health Differential in Finland”, Yearbook of Population Research in Finland, 41, 61-73.

SAHI T. (1974), “Lactose Malabsorption in Finnish Speaking and Swedish Speaking Populations in Finland”, Scandinavian Journal of Gastroenterology, 9, 303-308.

SAULI H. (1979), Ammatti ja kuolleisuus 1971-75, Research Reports, No. 54. Helsinki, Statistics Finland.

SCHNEIDER D., GREENBERG M.R., LU L.L. (1997a), “Early Life Experiences Linked to Diabetes Mellitus. A Study of African-American Migration”, Journal of the National Medical Association, 89, 29-34.

SCHNEIDER D., GREENBERG M.R., LU L.L. (1997b), “Region of Birth and Mortality from Circulatory Diseases among Black Americans”, American Journal of Public Health, 87, 800-804.

STATISTICS FINLAND (1991), Life Tables. Population 1991, No. 15. Helsinki, Statistics Finland.

STATISTICS FINLAND (1997), Life Tables. Population 1997, No. 4. Helsinki, Statistics Finland.

STATISTICS FINLAND (2004), Vital Statistics. Population 2004, No. 9. Helsinki, Statistics Finland.

STATISTICS FINLAND (2005a), Terveys, Kuolemansyyt, http://statfin.stat.fi/statweb/start.asp?LA=fi&DM=SLFI&lp=catalog&clg=terveys

STATISTICS FINLAND (2005b), Maailma numeroina: Elinajanodote, http://www.stat.fi/tup/maanum/hakemisto.html

STATISTICS FINLAND (2005c). Rekisteriseloste: Väestölaskentojen pitkittäistiedosto 1970-1995, http://www.stat.fi/meta/rekisteriselosteet/rekisteriseloste_vaestolaskenta70-95.html

STRACHAN D.P., LEON D.A., DODGEON B. (1995), “Mortality from Cardiovascular Disease among Interregional Migrants in England and Wales”, British Medical Journal, 310, 423-427.

STRENG O. (1935), “Die Blutgruppenforschung in der Anthropologie”, Acta Societatis Medicorum Fennicae Duodecim, 17 A (17/8), 1-318.

STRENG O. (1937), “Blutgruppenforschung und Anthropologie”, Zeitschrift für Rassenphysiologie, 9, 97-111.

VALKONEN T. (1987), “Male Mortality from Ischaemic Heart Disease in Finland: Relation to Region of Birth and Region of Residence”, European Journal of Population, 3, 61-83.

VALKONEN T., KAUPPINEN, T.M. (2001), “Miesten kuolleisuuden alue-erot ja sosiaalinen segregaatio pääkupunkiseudulla”, Kvartti, 1, 7-21.

Page 35: Regional mortality variation in Finland: a study of two

REGIONAL MORTALITY VARIATION IN FINLAND…

203

VALKONEN T., MARTELIN T., RIMPELÄ A., NOTKOLA V., SAVELA S. (1992), Sosioekonomiset kuolleisuuserot 1981-90, SVT Väestö, No. 8. Helsinki, Central Statistical Office of Finland.

VALKONEN T., MARTIKAINEN P. (1990), “Sepelvaltimotautikuolleisuuden väestöryhmittäinen kehitys Suomessa”, Sosiaalilääketieteellinen Aikakauslehti, 27, 273-288.

VIRTARANTA-KNOWLES K., SISTONEN P., NEVANLINNA H.R. (1991), “A Population Genetic Study in Finland: Comparisons of the Finnish- and Swedish-speaking Populations”, Human Heredity, 41, 248-264.

WORKMAN P.L., MIELKE J.H., NEVANLINNA H.R. (1976), “The Genetic Structure of Finland”, American Journal of Physical Anthropology, 44, 341-367.

Page 36: Regional mortality variation in Finland: a study of two

App

endi

x

Tab

le A

1 –

Rel

ativ

e m

orta

lity

of F

inni

sh-s

peak

ing

men

by

regi

on o

f res

iden

ce a

nd b

irth

reg

ion,

adj

uste

d fo

r ag

e, ti

me,

ed

ucat

iona

l lev

el a

nd fa

mily

type

B

irth

regi

on

Reg

ion

of

resi

denc

e T

urku

&

Por

i H

elsi

nki

area

U

usim

aa

Vaa

sa

Häm

e K

ymi

Mik

keli

Cen

tral

Fi

nlan

d K

uopi

o N

orth

K

arel

ia

Oul

u L

apla

nd

Ced

ed

Kar

elia

In

- m

igra

ted

Tur

ku &

Po

ri

100

102

107

104

110

113

127

128

127

117

116

116

109

110

Hel

sink

i ar

ea

108

124

112

112

109

118

113

118

125

133

126

117

123

117

Uus

imaa

10

1 12

0 10

3 11

3 11

0 11

4 12

0 13

5 11

8 11

4 13

0 15

9 11

0 11

3 V

aasa

10

6 9

2 9

5 10

2 11

4 13

0 12

9 11

0 13

9 12

2 10

8 12

8 11

9 11

6 H

äme

108

105

103

108

106

115

114

115

118

123

118

137

114

113

Kym

i 10

4 12

6 11

4 10

5 11

5 11

1 11

7 12

3 12

6 11

8 12

1 16

5 11

4 11

5 M

ikke

li 9

6 12

6 12

5 9

9 12

1 11

0 11

3 11

8 12

1 12

9 11

5 16

5 11

7 11

7 C

entr

al

Finl

and

94

115

94

108

104

119

114

110

119

126

114

127

121

116

Kuo

pio

109

93

93

113

123

117

113

117

116

115

100

158

115

114

Nor

th

Kar

elia

10

0 12

6 9

1 9

3 9

7 11

7 11

5 10

1 12

8 11

8 11

4 12

8 12

4 12

1

Oul

u 99

13

4 10

5 10

9 10

9 12

2 12

9 11

0 12

5 11

0 11

1 11

0 11

4 11

4 L

apla

nd

108

99

144

117

115

127

115

110

122

112

119

108

113

115

Out

-m

igra

ted

106

112

109

109

110

116

116

118

123

122

118

123

115

N

ote:

bor

n an

d re

sidi

ng in

Tur

ku &

Por

i=10

0.

204

Page 37: Regional mortality variation in Finland: a study of two

Tab

le A

2 –

Rel

ativ

e m

orta

lity

of F

inni

sh-s

peak

ing

wom

en b

y re

gion

of r

esid

ence

and

bir

th r

egio

n, a

djus

ted

for

age,

tim

e,

educ

atio

nal l

evel

and

fam

ily ty

pe

Birt

h re

gion

R

egio

n of

re

side

nce

Tur

ku

& P

ori

Hel

sink

i ar

ea

Uus

imaa

V

aasa

H

äme

Kym

i M

ikke

li C

entr

al

Finl

and

Kuo

pio

Nor

th

Kar

elia

O

ulu

Lap

land

C

eded

K

arel

ia

In-

mig

rate

d

T

urku

&

Pori

10

0 11

3 10

4 9

9 9

8 11

9 11

0 10

6 11

7 11

7 10

7 12

0 10

7 10

7

Hel

sink

i ar

ea

98

112

101

103

97

107

108

102

113

115

109

103

112

106

Uus

imaa

10

0 11

0 10

3 10

4 10

6 11

0 11

3 12

5 11

6 12

0 11

7 13

9 11

5 11

2 V

aasa

9

6 11

2 11

4 10

4 10

7 10

6 10

5 10

3 11

9 11

4 10

8 11

2 12

0 11

2 H

äme

100

107

103

101

103

111

106

110

119

115

111

110

111

109

Kym

i 10

4 12

2 9

5 10

6 10

8 11

6 11

8 11

6 12

4 12

1 10

6 9

4 11

7 11

5 M

ikke

li 10

4 12

4 9

8 9

9 10

7 11

4 11

8 12

9 11

5 12

4 11

2 12

9 11

7 11

6 C

entr

al

Finl

and

111

100

114

102

110

112

116

116

117

134

109

108

120

116

Kuo

pio

99

142

108

108

104

114

115

115

115

112

103

113

115

114

Nor

th

Kar

elia

9

6 11

0 9

5 10

3 10

1 12

6 10

2 9

9 12

4 12

0 9

8 7

4 12

7 12

1

Oul

u 11

7 9

1 10

0 10

6 9

4 11

5 12

8 10

6 11

1 11

9 11

2 10

8 11

3 11

1 L

apla

nd

84

118

128

99

113

141

127

123

129

140

121

112

115

117

Out

-m

igra

ted

100

111

101

102

101

111

111

109

116

118

111

109

113

N

ote:

bor

n an

d re

sidi

ng in

Tur

ku &

Por

i=10

0.

205

Page 38: Regional mortality variation in Finland: a study of two

Tab

le A

3 –

Rel

ativ

e m

orta

lity

of S

wed

ish-

spea

king

men

by

regi

on o

f res

iden

ce a

nd b

irth

reg

ion,

adj

uste

d fo

r ag

e, ti

me,

ed

ucat

iona

l lev

el a

nd fa

mily

type

B

irth

regi

on

Reg

ion

of

resi

denc

e T

urun

maa

Å

land

H

elsi

nki

area

W

este

rn

Uus

imaa

E

aste

rn

Uus

imaa

So

uthe

rn

Pohj

anm

aa

Cen

tral

Po

hjan

maa

N

orth

ern

Pohj

anm

aa

All

othe

r ar

eas

In-

mig

rate

d

T

urun

maa

10

0 8

0 11

8 10

7 13

5 11

4 10

9 11

1 10

6 10

8 Å

land

10

7 9

0 9

3 8

8 9

6 14

0 4

1 17

0 11

8 10

9 H

elsi

nki

area

10

6 10

5 11

1 10

1 10

6 11

1 10

9 9

9 10

6 10

4

Wes

tern

U

usim

aa

99

54

100

96

107

107

123

62

106

101

Eas

tern

U

usim

aa

99

171

110

109

106

81

141

119

92

103

Sout

hern

Po

hjan

maa

13

3 9

6 15

8 8

5 13

2 8

4 9

3 11

3 14

1 11

1

Cen

tral

Po

hjan

maa

10

2 7

0 9

4 7

8 11

5 9

4 9

5 9

0 9

4 9

3

Nor

ther

n Po

hjan

maa

18

7 8

0 11

5 15

4 24

1 10

0 7

7 9

3 8

9 9

2

All

othe

r ar

eas

99

77

102

82

94

177

104

92

100

97

Out

-m

igra

ted

104

89

104

100

107

104

98

96

101

N

ote:

bor

n an

d re

sidi

ng in

Tur

unm

aa=1

00.

206

Page 39: Regional mortality variation in Finland: a study of two

Tab

le A

4 –

Rel

ativ

e m

orta

lity

of S

wed

ish-

spea

king

wom

en b

y re

gion

of r

esid

ence

and

bir

th r

egio

n, a

djus

ted

for

age,

tim

e,

educ

atio

nal l

evel

and

fam

ily ty

pe

Birt

h re

gion

R

egio

n of

re

side

nce

Tur

unm

aa

Åla

nd

Hel

sink

i ar

ea

Wes

tern

U

usim

aa

Eas

tern

U

usim

aa

Sout

hern

Po

hjan

maa

C

entr

al

Pohj

anm

aa

Nor

ther

n Po

hjan

maa

A

ll ot

her

area

s In

- m

igra

ted

Tur

unm

aa

100

91

101

100

104

101

104

88

98

99

Åla

nd

106

80

104

88

75

56

187

56

95

95

Hel

sink

i ar

ea

107

105

103

94

102

97

95

101

101

100

Wes

tern

U

usim

aa

103

133

107

97

104

68

115

113

112

108

Eas

tern

U

usim

aa

89

107

81

97

102

101

101

127

105

96

Sout

hern

Po

hjan

maa

19

5 3

9 12

3 8

9 12

3 9

5 14

2 7

9 10

4 11

7

Cen

tral

Po

hjan

maa

10

6 8

0 7

6 8

8 9

1 7

8 9

6 8

6 11

6 9

3

Nor

ther

n Po

hjan

maa

8

9 8

9 9

6 8

8 9

4 7

2 11

6 9

6 9

0 9

3

All

othe

r ar

eas

91

98

90

94

102

126

91

99

97

94

Out

-m

igra

ted

103

100

94

94

102

82

107

95

100

N

ote:

bor

n an

d re

sidi

ng in

Tur

unm

aa=1

00.

207

Page 40: Regional mortality variation in Finland: a study of two

T

able

A5

– N

umbe

r of

obs

erva

tions

in o

ne-h

undr

eds,

by

regi

on o

f res

iden

ce a

nd b

irth

reg

ion,

Fin

nish

-spe

akin

g m

en

Birt

h re

gion

R

egio

n of

re

side

nce

Tur

ku

& P

ori

Hel

sink

i ar

ea

Uus

imaa

V

aasa

H

äme

Kym

i M

ikke

li C

entr

al

Finl

and

Kuo

pio

Nor

th

Kar

elia

O

ulu

Lap

land

C

eded

K

arel

ia

In-

mig

rate

d

T

urku

&

Pori

13

80

17

20

39

82

21

14

15

16

12

23

5 57

9 84

3

Hel

sink

i ar

ea

114

296

122

67

179

79

74

54

69

48

45

11

392

1254

Uus

imaa

40

32

26

7 25

66

28

19

16

20

15

17

3

212

493

Vaa

sa

15

4 3

901

15

7 4

13

5 3

17

3 79

16

7 H

äme

109

31

49

68

1127

54

50

64

32

28

33

6

595

1120

K

ymi

16

8 21

11

26

55

3 64

14

24

23

10

2

387

606

Mik

keli

7 5

5 6

14

32

518

14

25

15

5 1

130

256

Cen

tral

Fi

nlan

d 10

6

6 27

29

13

27

49

9 37

11

11

2

139

318

Kuo

pio

5 4

3 4

7 6

39

9 51

4 15

11

2

217

320

Nor

th

Kar

elia

3

2 1

3 5

9 13

3

17

406

6 1

144

207

Oul

u 9

4 3

31

12

8 7

10

33

9 87

4 16

14

2 28

6 L

apla

nd

7 2

2 19

9

5 5

5 13

6

68

296

109

249

Out

-m

igra

ted

335

117

234

300

443

261

316

216

290

185

244

51

3127

208

Page 41: Regional mortality variation in Finland: a study of two

T

able

A6

– N

umbe

r of

obs

erva

tions

in o

ne-h

undr

eds,

by

regi

on o

f res

iden

ce a

nd b

irth

reg

ion,

Fin

nish

-spe

akin

g w

omen

B

irth

regi

on

Reg

ion

of

resi

denc

e T

urku

&

Por

i H

elsi

nki

area

U

usim

aa

Vaa

sa

Häm

e K

ymi

Mik

keli

Cen

tral

Fi

nlan

d K

uopi

o N

orth

K

arel

ia

Oul

u L

apla

nd

Ced

ed

Kar

elia

In

- m

igra

ted

Tur

ku &

Po

ri

2118

25

36

80

16

4 43

36

34

41

31

62

10

92

4 14

87

Hel

sink

i ar

ea

187

463

226

118

348

166

196

137

188

133

116

27

787

2629

Uus

imaa

58

44

36

9 42

11

0 53

47

35

47

35

40

6

336

852

Vaa

sa

26

6 5

1340

28

13

8

31

10

8 42

8

137

321

Häm

e 19

5 48

86

13

0 17

34

104

104

136

71

62

80

16

1009

20

40

Kym

i 19

13

39

18

42

83

2 12

1 26

48

45

22

4

635

1030

M

ikke

li 8

9 7

10

23

52

728

26

48

30

13

3 21

8 44

7 C

entr

al

Finl

and

17

9 9

43

44

22

45

697

63

18

23

4 22

4 52

2

Kuo

pio

7 6

4 9

11

15

70

17

777

32

27

4 33

0 53

1 N

orth

K

arel

ia

4 3

2 5

6 15

22

5

29

577

12

3 23

1 33

6

Oul

u 11

5

5 47

16

14

11

16

51

17

12

50

33

206

433

Lap

land

7

2 2

19

8 5

6 6

16

7 10

8 40

8 14

3 33

1 O

ut-

mig

rate

d 54

0 16

9 42

0 52

1 80

1 50

3 66

5 46

9 61

2 41

8 54

5 11

6 51

81

209

Page 42: Regional mortality variation in Finland: a study of two

T

able

A7

– N

umbe

r of

obs

erva

tions

in o

ne-h

undr

eds,

by

regi

on o

f res

iden

ce a

nd b

irth

reg

ion,

Sw

edis

h-sp

eaki

ng m

en

Birt

h re

gion

R

egio

n of

re

side

nce

Tur

unm

aa

Åla

nd

Hel

sink

i ar

ea

Wes

tern

U

usim

aa

Eas

tern

U

usim

aa

Sout

hern

Po

hjan

maa

C

entr

al

Pohj

anm

aa

Nor

ther

n Po

hjan

maa

A

ll ot

her

area

s In

- m

igra

ted

Tur

unm

aa

109

2 5

3 2

1 1

1 17

32

Å

land

4

77

1 1

1 1

0 0

7 16

H

elsi

nki

area

23

3

165

48

35

2 6

5 70

19

3

Wes

tern

U

usim

aa

9 1

16

136

4 0

1 1

17

50

Eas

tern

U

usim

aa

2 0

9 5

136

0 1

1 10

29

Sout

hern

Po

hjan

maa

0

0 0

0 0

121

2 0

1 4

Cen

tral

Po

hjan

maa

2

0 2

1 1

8 13

5 4

5 23

Nor

ther

n Po

hjan

maa

1

0 1

1 1

1 4

130

40

48

All

othe

r ar

eas

7 1

13

6 6

1 2

2 20

37

Out

-m

igra

ted

48

8 47

66

50

14

18

14

16

7

210

Page 43: Regional mortality variation in Finland: a study of two

T

able

A8

– N

umbe

r of

obs

erva

tions

in o

ne-h

undr

eds,

by

regi

on o

f res

iden

ce a

nd b

irth

reg

ion,

Sw

edis

h-sp

eaki

ng w

omen

B

irth

regi

on

Reg

ion

of

resi

denc

e T

urun

maa

Å

land

H

elsi

nki

area

W

este

rn

Uus

imaa

E

aste

rn

Uus

imaa

So

uthe

rn

Pohj

anm

aa

Cen

tral

Po

hjan

maa

N

orth

ern

Pohj

anm

aa

All

othe

r ar

eas

In-

mig

rate

d

T

urun

maa

16

3 3

8 6

3 1

2 1

32

56

Åla

nd

7 10

1 3

2 1

2 1

1 16

32

H

elsi

nki

area

39

5

302

91

73

4 12

8

158

391

Wes

tern

U

usim

aa

14

1 22

18

7 7

1 2

1 33

83

E

aste

rn

Uus

imaa

4

1 16

8

204

1 1

1 22

55

So

uthe

rn

Pohj

anm

aa

0 0

1 0

1 16

6 3

1 2

9 C

entr

al

Pohj

anm

aa

2 0

3 2

1 14

19

8 5

13

41

Nor

ther

n Po

hjan

maa

2

0 3

2 1

1 8

180

58

75

All

othe

r ar

eas

8 1

16

8 11

1

3 2

34

50

Out

-m

igra

ted

77

13

72

118

98

26

33

22

334

211