Moving on: How depressive symptoms,social support,and health behaviors predict residential mobility

Abstract

Background:There are major health inequalities between residential areas. However, it remains unclear whether these inequalities are due to social causation or selective residential mobility, because little is known about the associations between health-related factors and selective residential mobility. This study examined how depressive symptoms, social support, and health behaviors are associated with subsequent residential mobility, as measured by frequency, distance, and direction of moves. Methods: Participants were selected from the Young Finns prospective cohort study (N = 3017) with four study waves in 1992, 1997, 2001, and 2007. Complete residential mobility history was gathered for each participant from registry data. Residential mobility was assessed over three-year periods following each study wave. The direction of mobility was measured as changes in residential location’s population density, health index, mortality index, and unemployment. Area characteristics were used as covariates together with the participants’ age, sex, and education. Results: Individuals reporting higher social support from friends were more likely to move (b = 0.20; 95% confidence interval (CI): 0.08, 0.37) and move more frequently (b = 0.08; 95% CI: 0.01, 0.14), whereas individuals reporting higher social support from their family members were less likely to move (b = −0.08; 95% CI: −0.14, −0.02). Better health behaviors were associated with longer moving distances (b = 0.14; 95% CI: 0.06, −0.23). None of the individual characteristics were associated with the direction of moves as measured by changes in municipality characteristics. Conclusions: Some of the health-related characteristics are associated with residential mobility in Finland. However, these individual characteristics do not seem to predict systematic selective residential mobility across municipalities with different regional health profiles.

Keywords

Selective residential mobility social support health behavior health inequality longitudinal

Introduction

There are major health inequalities between residential areas [1,2]. Whilst studies of neighborhood effects have implied that adverse neighborhood characteristics may cause poor physical and mental health and health behaviors [3,4], it is also possible that people with poorer health tend to live in adverse neighborhoods because they have fewer opportunities to move to better ones. Neighborhood-effect studies suggest that the social inequalities between neighborhoods may be partly caused by selective residential mobility [5 –7]. Other studies have reported similar findings on health-related residential mobility [8 –10]. More detailed data on health-related factors and residential mobility are needed to evaluate the contributions of social causation and selective residential mobility to the development of regional health inequalities.

In an Australian study of middle-aged women, poorer health was associated with a higher likelihood of moving, and smokers were more likely to move than non-smokers [8]. Likewise, individuals suffering from serious mental health problems such as schizophrenia have been shown to move more frequently [11,12]. Frequent residential mobility in itself has been associated with various social and health problems ranging from unemployment to mortality [13 –15]. People suffering from chronic illnesses are likely to seek locations that offer appropriate health services [8]. Furthermore, such illnesses can affect other aspects of life and have severe impact on an individual’s financial situation, which in turn limits the possibilities for relocating [16]. However, evidence that migrants have poor health come mainly from studies examining specific groups of people [8,9,15], rather than from population-based studies [17].

In a Finnish population-based longitudinal study, we examined whether depressive symptoms and health behaviors were associated with residential mobility between Finnish municipalities in participants aged 15–45 years. In addition, social support has been shown to buffer against physical and mental illnesses, including depression [18 –20]. Therefore, we also included social support as an indicator of health. We focused on the frequency, distance, and direction of residential mobility. The direction of residential mobility was used to measure selective mobility, determined as the difference between characteristics of in-migration and out-migration municipalities. Based on previous studies, we expected poor health behaviors to increase the frequency and distance of residential mobility [8]. Also, based on our earlier study from the same data, we anticipated social support to increase residential mobility between urban and rural municipalities, and that depressive symptoms would not to be associated with residential mobility [21].

Methods

Participants

The participants were 3017 individuals (54% women) from the ongoing Young Finns prospective cohort study [22]. The original Young Finns sample consisted of 3596 healthy Finnish children and adolescents aged 3, 6, 9, 12, 15, and 18 years at baseline in 1980. To be broadly representative of the Finnish population, the sample was gathered from and around five Finnish university cities with a medical school (Helsinki, Kuopio, Tampere, Turku, and Oulu). Participants were selected randomly based on their social security number. The study began in 1980 and eight follow-up study waves (1983, 1986, 1989, 1992, 1997, 2001, 2007, and 2012) have been completed since. The study has been approved by local ethics committees. In the current study, we used data from waves 1992, 1997, 2001, and 2007, and included those participants who had data for all individual characteristics (depressive symptoms, social support, and health behavior) on at least one of the study waves. From waves 1992, 1997, 2001, and 2007, data was available for 2322, 2091, 2066, and 2024 participants, respectively.

Municipality data were gathered from SOTKAnet [23], which contains statistical information on welfare and health in municipalities in Finland. In 2012, Finland was divided into 336 municipalities of different sizes (6–15,053 km² in land area). The median number of residents per municipality in 2012 was 5878 and the median population density was 10.90 persons/km². Participants included in the present study live in 331 unique municipalities.

Measurements

Residential mobility

The Population Register Centre of Finland provided a complete history of residential mobility up to the year 2013 for each participant. The history included the date and accurate coordinates of each move for each participant. The number of moves by a participant was derived from the data by counting the moves in the following three years after each study wave. As the exact measurement date varied between participants, the beginning of the year of a study wave was used as the starting date and the end of year three, as the end date (e.g. 1 January 1992–31 December 1995). As such, we had four counts of moves, one for each study wave interval. The coordinates were used to calculate the distance of each move during the three years following each study wave. The distance of moves was then categorized into five categories as follows: 1 = less than 5 km; 2 = 5–20 km; 3 = 20–50 km; 4 = 50–100 km; and 5 = over 100 km.

Depressive symptoms

A modified version of Beck’s Depression Inventory (BDI) was used to assess the depressive symptoms of the participants [24]. The symptoms were assessed in the study waves 1992, 1997, 2001, and 2007. The original BDI consists of 21 items with four alternative statements for each item. The modified version (mBDI) uses the second mildest statement of each original item, which are answered on a 5-point Likert scale (1=totally disagree, 5=totally agree). The mBDI was selected for use in the current study, because it has been suggested that it captures the depressive tendencies of a non-clinical population more efficiently than the original BDI [25]. The internal consistency of the measure was examined using Cronbach’s alphas, which indicates how well different items of the scale measure the same underlying construct. Values of the Cronbach alpha range between 0 and 1, and values higher than 0.70 are deemed acceptable [26]. Cronbach’s alphas for the modified version of the inventory were 0.88, 0.91, 0.92, 0.93, and 0.93, respectively for each study wave.

Social support

Social support was assessed using the Multidimensional Scale of Perceived Social Support [27] in 1992, 1997, 2001, and 2007. The measurement scale was divided into three subcategories, i.e., social support by family, social support by friends, and social support by significant other. Each subcategory had four items, which were rated on a 5-point Likert scale (1=totally disagree, 5=totally agree). Items included were, for example, “My friends really support me when I need support”, “I get emotional help and support I need from my family”, and “I have a special person who comforts me”. As a portion of our participants were 15 years of age in the first study wave, the “significant other” subcategory was translated as “close friend” and that translation was used for all the study waves. In the current study, the subcategories “support by friend” and “support by significant other” were combined. Cronbach’s alphas for the “social support by family” variable for each study wave were 0.90, 0.92, 0.92, and 0.94. For “support by friends” the alphas were 0.89, 0.91, 0.92, and 0.93, and for “significant other” 0.95, 0.95, 0.96, and 0.96. For the purpose of the study, the scales for “support by friend” and “support by significant other” were combined and divided by 2 to match the scale of the “social support by family” variable.

Health behavior index

Participants’ health behaviors were assessed using self-report questionnaires. Information was collected on the consumption of alcohol, smoking, frequency of exercise, and self-rated interest in health in study waves 2001 and 2007. Alcohol consumption was rated on a 6-point scale on the question “How often do you drink 6 units of alcohol or more?” (1 = less than twice a year or never; 2 = 2–6 times a year; 3 = Once a month; 4 = 2–3 times a month; 5 = once a week; and 6 = twice a week or more often). Smoking was reported on 5-point scale (1 = I have never smoked; 2 = I have quit /I am on a break; 3 = less than once a week; 4 = once a week or more, but not daily; and 5= one or more cigarettes a day). Frequency of exercising was rated on a 6-point scale (1 = never; 2 = once a month; 3 = once a week; 4 = 2–3 times a week; 5 = 4–6 times a week; and 6=daily). Self-rated interest in health was reported on a 5-point scale (1 = I barely pay any attention to my health; 2 = I only pay a little attention to my health; 3 = Neither little nor a lot; 4 = I pay some attention to my health; and 5 = I pay a lot of attention to my health). A combined measure for health behavior was formed by dichotomizing the responses on the scales described above and summing the resulting scores together. Scales were dichotomized according to the responses as follows: alcohol 0 = 1–3, 1 = 4–6; smoking 0 = 1–2, 1 = 3–5; exercise 0 = 1–3, 1 = 4–6; and interest in health 0 = 1–2, 1 = 3–5. The final scale ranged from 0 to 4, with the higher value representing better health behaviors.

Area characteristics

Municipality urbanicity was measured as population density and socioeconomic status as tax revenue per capita. Both measures were available for all study waves included in the current study. To describe the health of residents in a municipality, we used a health index that includes seven different groups of diseases: cancer; coronary heart disease; cerebrovascular diseases; diseases of the musculoskeletal system; mental health problems; accidental injuries; and dementia. The index was compared to the national level of health (national index = 100). The prevalence of each disease group was weighted within the index. The age- and sex-adjusted index was available for study waves 2001 and 2007. For the mortality characteristic, we used an index describing the portion of mortalities within the municipality residents. The index was computed for all the municipalities of Finland and compared to the mortality rate of the nation (national index = 100). The mortality index was adjusted for age and sex, and was available for all study waves used. For unemployment, an index describing the portion of unemployed of the total work force was used. The total work force included all residents aged 15–64 years. The index was available for all study waves.

Covariates

Age, sex, and education were included as covariates in all analyses. The education variable was self-reported in 1992, 2001, and 2007, and it was divided into four categories: 1 = vocational upper secondary school degree or similar; 2 = polytechnic degree; 3 = university studies (no degree); and 4 = university degree.

Statistical analysis

Associations between individual characteristics (depressive symptoms, social support, and health behavior) and different aspects of residential mobility were examined using regression analyses. Multilevel logistic regression analysis was used to examine the association between individual characteristics and whether participants moved at all or not. As the number of moves by a participant was an over-dispersed count variable, multilevel negative binomial regression analysis was used to examine the association between individual characteristics with the number of moves. Finally, multilevel ordered logistic regression analysis was used to examine the association between individual characteristics and categorically coded moving distance.

As a preliminary analysis, we examined whether municipality characteristics were associated with whether participants moved or not. For the main analysis, we examined separately the associations between individual characteristics and residential mobility adjusting for area socioeconomic status, area population density, area health index, area mortality index, area unemployment, age, sex, and education. We also ran interaction analyses to see whether individual characteristics were associated with residential mobility differently in different municipalities.

Finally, we ran multilevel logistic regression analysis to examine the direction of moving according to the area characteristics used for municipalities. We created dummy variables for the moves based on whether the move was to a municipality with a lower or higher population density, socioeconomic status, health index, mortality index, or unemployment. We ran separate multilevel logistic regression analyses for comparing those who did not move to those who moved upward in the area characteristics, and those did not move to those who moved downward in the area characteristics. Additionally, we also compared those who moved upwards to those who moved downwards in the area characteristics.

Results

The descriptive statistics are shown in Table I. Of the 3017 participants (54% women), 1124 had returned all the data from each of the four study waves (735 from three, 644 from two, and 514 from one) and 267 of the participants did not move during the study period. The total number of moves made by an individual ranged from 1 to 21. Moves between municipalities accounted for 35% of all the moves.

Table I.

Descriptive statistics of the 10,203 person-observations from 3017 unique individuals from the Young Finns Study (1992–2007), and municipality-area-level measures.

Variable	Total no.	No. of persons	%	Mean (SD)	Within-person SD
Sex
Men	4267	1385	53.0
Women	5936	1632	47.0
Age				32.1 (8.7)	7.1
Education
Vocational	4557	1894	64.9
Polytechnic	729	426	10.4
University (no degree)	249	190	3.6
University degree	1487	671	21.2
Number of moves				0.8 (1.1)	0.9
Distance of moves
<5 km	5772	2352	57.8
5–20 km	2047	1289	20.5
20–50 km	760	547	7.6
50–100 km	480	360	4.9
>100 km	928	611	9.3
Depressive symptoms^a				2.1 (0.7)	0.3
Social support by family^a				4.2 (0.9)	0.5
Social support by friends^a				4.2 (0.9)	0.5
Health behavior^b				2.5 (1.2)	0.5
Municipality
Urbanization				485.0 (892.6)	445.3
SES				3115.3 (705.7)	361.5
Health index				106.1 (18.5)	5.8
Mortality index				100.0 (7.7)	5.3
Unemployment index				12.4 (4.8)	3.8

SD: standard deviation; SES: socioeconomic status.

For categorical variables, the values are the number of total person-observations, number of unique persons, and percentages calculated from person-observations. For continuous variables, the values are means, overall standard deviations, and within-person standard deviations.

Rated on a scale of 1–5.

Rated on a scale of 0–4.

Of the area characteristics, higher population density (odds ratio (OR) = 1.21; 95% confidence interval (CI) = 1.12–1.29) and higher socioeconomic status (OR = 1.31; 95% CI = 1.20–1.43) were associated with higher likelihood of moving. Higher mortality index was associated with lower likelihood of moving (OR = 0.92; 95% CI = 0.86−0.99). Municipality unemployment and health indexes were not associated with the likelihood of moving. Results for the associations between depressive symptoms, social support, and health behaviors with different aspects of residential mobility are shown in Table II. Whereas higher social support from friends was associated with a higher propensity of moving, depression, social support from family and health behaviors were not associated with moving propensity. Higher social support from family was associated with a lower number of moves, and higher social support from friends was associated with a higher number of moves. All these associations remained after additionally controlling for area-level characteristics. Depressive symptoms and health behaviors were not associated with the number of moves, but better health behaviors were associated with longer moving distances. For further sensitivity analyses, we ran the analyses for the associations between individual variables and moving distances using deciles of migration distances. The same associations were observed in these analyses (data not shown), suggesting that the results were not sensitive to specific cut-off values for migration distances. None of the associations between depressive symptoms, social support, or health behaviors with moving upwards or downwards in area characteristics were significant (Table III).

Table II.

Associations between depressive symptoms, social support, health behaviors and moving, number of moves, and moving distance.

	Move/No move		Number of moves		Distance of moves
	b	95% CI	b	95% CI	b	95% CI
Model 1: Adjusted for age, sex, and education.
Depressive symptoms	0.02	(−0.09; 0.13)	0.04	(−0.02; 0.10)	−0.15	(−0.31; 0.01)
Social support by family	−0.02	(−0.10; 0.06)	−0.03	(−0.08; 0.01)	−0.04	(−0.16; 0.08)
Social support by friends	0.28	(0.20; 0.37)	0.16	(0.11; 0.21)	0.02	(−0.11; 0.15)
Health behaviors	−0.05	(−0.12; 0.03)	−0.03	(−0.08; 0.01)	0.14	(0.06; 0.23)
Model 2: Additionally adjusted for density of population, municipality SES, unemployment, health index, and mortality index.
Depressive symptoms	−0.01	(−0.15; 0.13)	0.02	(−0.06; 0.10)	−0.15	(−0.33; 0.02)
Social support by family	−0.08	(−0.19; 0.03)	−0.08	(−0.14; −0.02)	−0.11	(−0.26; 0.03)
Social support by friends	0.20	(0.08; 0.31)	0.08	(0.01; 0.14)	0.03	(−0.12; 0.18)
Health behaviors	−0.07	(−0.14; 0.01)	−0.03	(−0.07; 0.01)	0.14	(0.04; 0.24)

Regression coefficients and 95% CIs from logistic regression (Move/No move), negative binomial regression (Number of moves), and ordinal logistic regression (Distance of moves). Statistically significant b-values and confidence intervals are in bold.

Table III.

Associations between depression, social support, health behaviors, and direction of moving, as characterized by area variables.

	Population density		Area SES		Area health index		Area mortality index		Area unemployment
	OR	95% CI	OR	95% CI	OR	95% CI	OR	95% CI	OR	95% CI
Up movers vs. stayers	(N = 3030; 1503)		(N = 3006; 1502)		(N = 973; 645)		(N = 3022; 1508)		(N = 2985; 1500)
Depression	0.81	(0.57; 1.31)	0.97	(0.52; 1.33)	1.30	(0.62; 1.36)	1.00	(0.53; 1.31)	1.23	(0.64; 1.47)
Social support by family	1.40	(0.62; 1.16)	1.19	(0.57; 1.11)	0.77	(0.6; 1.06)	0.78	(0.55; 1.03)	0.82	(0.59; 1.07)
Social support by friends	0.85	(0.71; 1.3)	0.68	(0.62; 1.22)	0.69	(0.69; 1.21)	0.63	(0.64; 1.19)	0.81	(0.67; 1.21)
Health behaviors	1.13	(0.73; 1.12)	0.99	(0.77; 1.21)	0.87	(0.68; 1.03)	1.04	(0.71; 1.16)	1.00	(0.68; 1.08)
Down movers vs. stayers	(N = 3003; 1528)		(N = 3031; 1534)		(N = 966; 643)		(N = 3012; 1526)		(N = 3054; 1539)
Depression	0.94	(0.56; 1.42)	0.92	(0.47; 1.32)	0.99	(0.52; 1.24)	0.92	(0.57; 1.38)	1.05	(0.49; 1.2)
Social support by family	0.89	(0.53; 1.05)	0.84	(0.54; 1.15)	0.83	(0.6; 1.15)	0.79	(0.59; 1.15)	0.83	(0.55; 1.07)
Social support by friends	1.01	(0.7; 1.47)	0.92	(0.74; 1.66)	0.95	(0.75; 1.52)	0.91	(0.74; 1.54)	0.94	(0.76; 1.52)
Health behaviors	0.93	(0.64; 1.03)	0.99	(0.62; 1.07)	0.86	(0.76; 1.2)	0.93	(0.64; 1.06)	0.88	(0.75; 1.14)
Up movers vs. down movers	(N = 1103; 731)		(N = 1099; 728)		(N = 363; 273)		(N = 1102; 730)		(N = 1097; 727)
Depression	0.81	(0.32; 1.3)	0.97	(0.31; 1.62)	1.30	(0.57; 2.03)	1.00	(0.36; 1.63)	1.23	(0.52; 1.93)
Social support by family	1.40	(0.71; 2.09)	1.19	(0.54; 1.85)	0.77	(0.44; 1.11)	0.78	(0.4; 1.16)	0.82	(0.47; 1.17)
Social support by friends	0.85	(0.47; 1.23)	0.68	(0.3; 1.06)	0.69	(0.39; 0.99)	0.63	(0.31; 0.96)	0.81	(0.45; 1.18)
Health behaviors	1.13	(0.78; 1.49)	0.99	(0.64; 1.35)	0.87	(0.62; 1.11)	1.04	(0.68; 1.39)	1.00	(0.68; 1.31)

SES: socioeconomic status.

ORs and 95% CIs from logistic regression analysis. Adjust for age, sex, education, density of population, municipality SES, unemployment, health index, and mortality index. Up movers refers to participants who moved to either a municipality with higher population density, higher SES, better health index, higher mortality index, or higher unemployment index. Down movers refers to the opposite move on each scale. The number of observations in each regression analysis is shown as the total number of person-observations and the number of unique individuals.

Of the 60 potential interactions (4 individual characteristics × 5 municipality characteristics × 3 residential mobility outcomes), only depressive symptoms and social support with municipality unemployment were associated with distance of moving (online Supplementary Table 1). To clarify the direction of the interactions, we ran simple slope analyses (online Supplementary Figure 1). Those with higher depressive symptoms made shorter moves in low unemployment municipalities and longer moves in high unemployment municipalities. Those who received more support from friends or family made shorter moves in high unemployment municipalities and longer moves in low unemployment municipalities. However, these interaction effects need to be interpreted with caution because we tested for 60 interactions without specific hypotheses, and statistically significant interactions can be found by chance with multiple testing.

Discussion

The current study suggests that those who receive more social support from their friends are more likely to move, and move more often than those who receive less support from their friends. By contrast, those who receive more social support from their family tend to move less often than those who do not receive as much support from their family. Additionally, individuals with optimal health behaviors move longer distances than those whose health behaviors were less optimal. Depressive symptoms were not associated with residential mobility. Surprisingly, none of the health characteristics were associated with selective residential mobility with respect to the regional health profiles of municipalities.

The source of social support affected whether or not people moved and how often. A recent study found that individuals anticipating a mobile lifestyle in the near future were motivated to expand their social network [28]. Thus, it is plausible that the association between social support from friends and increased probability to move is partly explained by a preceding anticipation of a future move. Furthermore, as moving is an anxiety provoking life-event [29], a certain amount of social support might be needed for an individual to be able to cope with the stress that is associated with moving. It is also possible that a large social network of friends can create pull factors for moving. Having friends in other cities could potentially enable, for example, better employment chances. Those who mainly receive social support from their family might not have a wide enough social network in place to receive adequate social support after a move, and hence, choose not to move in the first place. Higher social support from family may also indicate closer ties to relatives more generally, and these social ties may decrease people’s willingness to leave their family members.

Contrary to our hypotheses, social support was not associated with residential mobility on the rural–urban continuum. In our earlier study [21], we found that people who moved to more urban areas received more social support from their friends than those living in rural areas. The evidence suggested that the difference could be equally explained by the fact that living in an urban environment increases social support from friends, and that people who receive a lot of support from their friends were likely to move to urban areas. Together with the present results, however, it seems more likely that urban living increases social support received from friends, as there was no evidence for selective mobility.

The Australian study of middle-aged women found that smokers moved longer distances than non-smokers [8]. We found that those with good health behaviors moved longer distances. The samples of the two studies are markedly different and thus, it is conceivable that the effects of health behaviors are different in a broader sample of a population. In order to get a more general view of the effects on residential mobility, we used an aggregate measure for health behaviors rather than looking at individual behaviors. As such, it is possible that any opposite effect smoking might have had on residential mobility was obscured by the effects of alcohol consumption, frequency of exercise, and self-rated interest in health. As expected, we did not find any association between depressive symptoms and residential mobility. Earlier studies have linked serious mental disorders to more frequent residential mobility [11,12]. However, participants in our study were relatively healthy. Thus, they would not have to deal with the negative consequences that a severe mental health issue might cause, such as the inability to work, which in turn might force people to move to a different residential area.

Two previous studies in Australia and the UK reported that health and health behaviors were associated with selective residential mobility across levels of neighborhood deprivation, so that individuals with poorer health were more likely to move to more deprived neighborhoods [5,6]. We found no evidence for selective residential mobility across municipalities with different levels of socioeconomic status, unemployment rate, population density, health index, or mortality rate. This suggests that health-related selective mobility is unlikely to create regional health inequalities in Finland, although our focus on moves across municipalities may have ignored selective residential mobility at a smaller scale. Two thirds of the moves in our sample were made within municipalities, and examining the direction of residential mobility at a zip-code area level could have yielded different results. In addition, the influence of health on residential mobility may depend on people’s life-course characteristics, such as marital status, parenthood, and employment status. Life-course characteristics have been associated with migration behavior [30], and age-dependent health inequalities between neighborhoods have been reported in the UK [31]. A detailed analysis of life-course dependent health selection was beyond the scope of our current analysis, but this topic should be investigated in future studies

Strengths and limitations

The fundamental strength of this study is the detailed residential mobility history of each participant based on registry data. Together with repeated measurements, these longitudinal data allowed us to examine residential mobility accurately in a population-based sample. Earlier studies have suffered from imprecise measures of residential mobility, such as annually recorded addresses [9,10,32,33], and self-reports, which are vulnerable to memory bias, especially for those who have moved numerous times [14,15]. However, the lack of proper zip-code area level data meant that we were unable to examine residential mobility at a smaller scale. As such, we might have missed any effects the individual characteristics could have had on residential mobility, for example, within cities. Furthermore, the results are likely to be generalizable only to countries within Europe due to cultural and demographic differences.

Conclusions

Selective residential mobility has been suggested as one of the reasons for growing inequalities between living areas. Our results suggest that while social support and individual health behaviors contribute to different aspects of mobility behavior, they are not associated with selective residential mobility between municipalities in Finland. Evidence on the topic at a smaller geographical scale could possibly illuminate the issue further.

Footnotes

Conflict of interest

None declared.

Funding

This work was financially supported by the Academy of Finland (L.K.J., grant numbers 258711 and 265869; M.J., grant number 268388; Elovainio, grant numbers 265977; T.L, grant number 286284), the Kone Foundation (M.J.) and the Juho Vainio Foundation (L.P-R.). The Young Finns Study has been financially supported by the Academy of Finland (grant numbers 126925, 121584, 124282, 129378 (Salve), 117787 (Gendi), and 41071 (Skidi)), the Social Insurance Institution of Finland, Kuopio, Tampere and Turku University Hospital Medical Funds (grant number X51001 for Dr Lehtimäki), Juho Vainio Foundation, Paavo Nurmi Foundation, Finnish Foundation of Cardiovascular Research (T.L), Yrjö Jahnsson Foundation (T.L.), and Finnish Cultural Foundation, Tampere Tuberculosis Foundation and Emil Aaltonen Foundation (for Dr Lehtimäki).

Supplemental material

Online supplementary data is available at .

References

Diez Roux

Mair

. Neighborhoods and health. Ann N Y Acad Sci 2010;1186:125–45.

Schulz

Israel

Williams

. Social inequalities, stressors and self reported health status among African American and white women in the Detroit metropolitan area. Soc Sci Med 2000;51(11):1639–53.

Ivory

Blakely

Richardson

. Do changes in neighborhood and household levels of smoking and deprivation result in changes in individual smoking behavior? A large-scale longitudinal study of New Zealand adults. Am J Epidemiol 2015;(14):431–40.

Ross

Mirowsky

. Neighborhood disadvantage, disorder, and health. J Health Soc Behav 2001;42(3):258–76.

Jokela

. Does neighbourhood deprivation cause poor health? Within-individual analysis of movers in a prospective cohort study. J Epidemiol Community Heal 2015;69(9):1–6.

Jokela

. Are neighborhood health associations causal? A 10-year prospective cohort study with repeated measurements. Am J Epidemiol 2014;180(8):776–84. Available from http://aje.oxfordjournals.org/cgi/doi/10.1093/aje/kwu233

Curtis

Setia

Quesnel-Vallee

. Socio-geographic mobility and health status: A longitudinal analysis using the National Population Health Survey of Canada. Soc Sci Med 2009;69(12):1845–53. Available from http://dx.doi.org/10.1016/j.socscimed.2009.08.004

Larson

Bell

Young

. Clarifying the relationships between health and residential mobility. Soc Sci Med 2004;59(10):2149–60.

Tunstall

Mitchell

Pearce

. The general and mental health of movers to more- and less-disadvantaged socio-economic and physical environments within the UK. Soc Sci Med 2014;118:97–107.

10.

Halliday

Kimmitt

. Selective migration and health in the USA, 1984–93. Popul Stud (NY) 2008;62(3):321–34.

11.

DeVerteuil

Hinds

Lix

. Mental health and the city: Intra-urban mobility among individuals with schizophrenia. Heal Place 2007;13(2):310–23.

12.

Lix

Hinds

DeVerteuil

. Residential mobility and severe mental illness: A population-based analysis. Adm Policy Ment Heal Ment Heal Serv Res 2006;33(2):160–71.

13.

Exeter

Sabel

Hanham

. Movers and stayers: The geography of residential mobility and CVD hospitalisations in Auckland, New Zealand. Soc Sci Med 2015;133:331–9.

14.

Oishi

Schimmack

. Residential mobility, well-being, and mortality. J Pers Soc Psychol 2010;98(6):980–94.

15.

Jelleyman

Spencer

. Residential mobility in childhood and health outcomes: A systematic review. J Epidemiol Community Health 2008;62(7):584–92.

16.

Rabe

Taylor

. Residential mobility, quality of neighbourhood and life course events. J R Stat Soc Ser A Stat Soc 2010;173:531–55.

17.

Champion

. Population movement within the UK. Focus People Migr 2005:91–114.

18.

Kawachi

Berkman

. Social ties and mental health. J Urban Health 2001;78(3):458–67.

19.

Cohen

Wills

. Stress, social support, and the buffering hypothesis. Psychol Bull 1985;98(2):310–57.

20.

Klineberg

Clark

Bhui

. Social support, ethnicity and mental health in adolescents. Soc Psychiatry Psychiatr Epidemiol 2006;41(9):755–60.

21.

Airaksinen

Hakulinen

Elovainio

. Neighborhood effects in depressive symptoms, social support, and mistrust: Longitudinal analysis with repeated measurements. Soc Sci Med 2015;136–137:10–6.

22.

Raitakari

Juonala

Rönnemaa

. Cohort profile: The cardiovascular risk in young Finns study. Int J Epidemiol 2008;37(6):1220–6.

23.

SOTKAnet. National Institute for Health and Welfare. SOTKAnet statistics and indicator bank, http://www.sotkanet.fi (2015, accessed 3 January 2015).

24.

Katainen

Räikkönen

Keltikangas-Järvinen

. Adolescent temperament, perceived social support, and depressive tendencies as predictors of depressive tendencies in young adulthood. Eur J Pers 1999;13(April 1998):183–207.

25.

Rosenström

Jokela

Puttonen

. Pairwise measures of causal direction in the epidemiology of sleep problems and depression. PLoS One 2012;7(11):1–13.

26.

Tavakol

Dennick

. Making sense of Cronbach’s alpha. Int J Med Educ 2011;2:53–5.

27.

Blumenthal

Burg

Barefoot

. Social support, type A behavior, and coronary artery disease. Psychosom Med 1987;49(4):331–40.

28.

Oishi

Kesebir

Miao

. Residential mobility increases motivation to expand social network: But why? J Exp Soc Psychol 2013;49(2):217–23.

29.

Holmes

Rahe

. The Social Readjustment Rating Scale. J Psychosom Res 1967;11(5):213–8.

30.

Kley

. Explaining the Stages of migration within a life-course framework. Eur Sociol Rev 2011;27(4):469–86. Available from http://esr.oxfordjournals.org/cgi/doi/10.1093/esr/jcq020

31.

Norman

Boyle

. Are health inequalities between differently deprived areas evident at different ages? A longitudinal study of census records in England and Wales, 1991–2001. Health Place 2014;26:88–93. Available from http://linkinghub.elsevier.com/retrieve/pii/S1353829213001743

32.

Norman

Boyle

Rees

. Selective migration, health and deprivation: A longitudinal analysis. Soc Sci Med 2005;60(12):2755–71.

33.

Verheij

van de Mheen

de Bakker

. Urban-rural variations in health in The Netherlands: Does selective migration play a part? J Epidemiol Community Health 1998;52(8):487–93.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.07 MB