Implications of Social and Material Deprivation for Changes in Health of Older People

Abstract

Objective: We use the panel structure of the Survey of Health, Ageing and Retirement in Europe (SHARE) data for 14 countries to examine the implications of material and social deprivation for health deterioration in old age and mortality. Method: To minimize the potential endogeneity bias, we examine the relationship between deprivation and changes in health rather than levels of health. We include a substantial set of fixed “initial conditions,” and extend the controls with health measures, as observed at the initial period. Results: The results of the probit regression suggest a strong and statistically significant relationship between measures of material and social deprivation and changes in physical and mental health. Mortality is only affected by the social dimension of deprivation. Discussion: Treating material and social deprivation separately rather than as a single social exclusion indicator allows for more specific identification of the role of the two dimensions, which might be important for policy decisions.

Keywords

health in later life mortality material and social deprivation social exclusion

Introduction

Limiting the extent of poverty and social exclusion has been one of the key objectives at the national and European level for over two decades, and reducing the number of people at risk of poverty and social exclusion by 20 million became one of the targets of the European Commission’s ‘Europe 2020 strategy’. At the same time, despite the intuitive nature of the concept there has been an ongoing debate in the literature on how social exclusion should be defined and measured (see, for example, Walsh, Scharf, Cullinan, & Finn, 2012; Walsh, Scharf, & Keating, 2016). Given the importance of the social context in the understanding of the concepts of deprivation and exclusion, they are by definition ever changing. Thus, it is impossible to expect a final verdict on the approach to who is and who is not materially or socially deprived, and who can be classified as socially excluded. This, however, should not stop researchers from trying to identify convincing broad measures of well-being with a focus on deprivation or exclusion. One related concept was proposed in Myck, Najsztub, and Oczkowska (2015) where risk of social exclusion among people aged 50+ was measured with respect to the relative position of older individuals in the distribution of material and social deprivation indices. These indices were constructed using data from Wave 5 of the Survey of Health, Ageing and Retirement in Europe (SHARE) conducted in year 2013 and were developed with the use of questionnaire items specifically designed to identify aspects of well-being particularly relevant to individuals aged 50+, such as material conditions, quality of the neighborhood, and accessibility to services. Binary indicators signaling individual deprivation with respect to these items were employed with hedonic weights to compute material and social deprivation indices. We then constructed an indicator of severe deprivation by taking individuals allocated above the 75th percentile of the total distribution of each dimension of deprivation. This last measure sheds light on the population at risk of social exclusion. Further in the article, with the availability of data from Wave 6 of the SHARE survey collected in year 2015, all three measures of deprivation are used to analyze the implications of old age material and social deprivation for changes in health.

While measuring deprivation and social exclusion in a cross-sectional setting gives us important knowledge about the material and social state of the examined population, using this information for contemporaneous correlations is unlikely to be very informative of the role they play in determining other outcomes. From this perspective, there are a number of advantages in using the SHARE data. First, using the panel structure of SHARE gives us the possibility to look into how changes observed between two waves of the survey relate to the values of deprivation and the risk of social exclusion as identified in the initial wave. Second, the SHARE data contain a substantial set of information related to the life history of respondents which can be considered as correlates of the so-called “initial conditions.” This, in turn, to some extent allows us to further address the issue of endogeneity bias in our estimates resulting from unobservable factors.

In this article, we focus specifically on one of the key determinants of individuals’ well-being, namely health. Thus, the aim of the article can be presented as twofold. First, using complex measures of deprivation designed specifically for older individuals, we validate the hypothesis that deprivation, both material and social, contributes to the deterioration of health at older ages. Second, by examining the relation between the two, we are able to demonstrate that among older individuals the combination of material and social deprivation—our measure of social exclusion—is detrimental to physical and mental health.

Given the complex nature of the relationship between health and material and social well-being (see, for example, Cambois, Solé-Auró, & Robine, 2016; Deaton, 2008; Ettner, 1996; Haan & Myck, 2009; Marmot, 2002; Teerawichitchainan & Knodel, 2015; Testi & Ivaldi, 2009; Tubeuf & Jusot, 2011; Winkleby, Jatulis, Frank, & Fortmann, 1992), the approach we take is still unlikely to uncover a clear causal relationship between deprivation, social exclusion, and health outcomes. As we explain in a formal presentation of the model we estimate, there may still be unobservable factors which correlate both with the level of deprivation and exclusion and with changes in health, thus causing some bias in the estimates. However, in our approach, we examine the relationship between health and material and social well-being under very strict assumptions and—as our results show—we still find a strong and statistically significant relationship between changes in health and our key variables of interest.

In addition, even if the analyzed relationship cannot be given a clear causal interpretation, relating the information on the levels of material and social well-being with changes in health can be informative in at least two important and policy-relevant aspects. First, the extent of future health developments among the currently deprived or excluded population can serve as indicators with regard to the expected developments in the level of health of the population, and consequently as signals concerning future health care needs. Second, by examining the role of material and social deprivation separately, we can specify the relative role of each of these domains for developments in health at a later stage in people’s lives. This can be instrumental in designing specifically targeted policy interventions.

The Relationship Between Health and Deprivation

Literature Review

Since at least the 1970’s, a significant effort has been made to identify the determinants of different aspects of well-being (Gaertner, 1993; Nussbaum, 2001; Sen, 2000; Townsend, 1979), and it has become widely accepted that quality of life should not be perceived solely with reference to income and income poverty. Bhalla and Lapeyre (1997) noted that one possible way to go beyond the income and commodity notion of evaluating quality of life was through the concept of social exclusion. The concept itself is far from straightforward and its definition has been approached from many different directions, reflecting, in particular, its relative nature (Atkinson & Hills, 1998; Levitas et al., 2007; Sen, 2000). Another commonly accepted characteristic of social exclusion is multidimensionality. The number of dimensions in social exclusion and the range of life domains they might cover have been widely discussed in the literature, resulting in almost as many classifications as studies with reference to economic, social, and institutional dimensions (Pirani, 2013); financial difficulties, basic necessities, housing conditions, durables, health, social contact, and dissatisfaction (Chakravarty & D’Ambrosio, 2006); or material deprivation, social rights, social participation, and social integration (Jehoel-Gijsbers & Vrooman, 2008), to name just a few. Regardless of the classification, the factors common to all are deprivation in material and deprivation in social aspects of life. Among many others, a broad but straightforward definition of these dimensions of deprivation was proposed in Jehoel-Gijsbers and Vrooman (2008), which served as a foundation in Myck et al. (2015) and which we followed throughout this article. Material deprivation may be defined as deficiency in basic material needs, and social deprivation corresponds to social isolation and a lack of social support.

It has been well established in the literature that a lack of command over material resources can have significant adverse effects on health, leading to poor health outcomes as defined with both broader and more precise measures of health. Using simple self-assessment of health, which is both easy to understand and to implement in a social survey, Doebler and Glasgow (2017) analyze the relationship between health and material deprivation among people aged above 64 years. While Adena and Myck (2014) start off with similar self-reported declaration of health status, they extend their examination of health-related poverty implications with indicators of having difficulties with activities of daily life and being troubled by specific ill-health symptoms, such as breathlessness. Benach, Yasui, Borrell, Sáez, and Pasarin (2001) show strong detrimental associations of material deprivation and civilization diseases such as diabetes, heart diseases, or lung cancer, providing some further evidence that broadly defined material conditions are correlated with not only future health outcomes but also mortality, which has also been reported in other studies (e.g., Carstairs & Morris, 1989). Rohde, Tang, Osberg, and Rao (2017) show that even the prospects of economic losses could lead to worsened health.

The diversity of definitions of material hardship cited above suggests that the way we measure material well-being has important implications for the identified relationship with health. A number of studies have shown, for example, that changes in health are generally less responsive to income-based measures of poverty compared with more general concepts such as measures of (relative) deprivation or self-assessed material conditions (Adena & Myck, 2014; Gero, Kondo, Kondo, Shirai, & Kawachi, 2017; Gunasekara, Carter, Crampton, & Blakely, 2013). This has been explained by the fact that broader definitions of material hardship are closer in the chain of events leading to bad health than income. Moreover, in particular with regard to older people, income-based measures may be poor indicators of overall material conditions, as they do not account for the level of available assets and they are insensitive to varying needs of individuals conditional on such characteristics as health or disability (Detollenaere, Desmarest, Boeckxstaens, & Willems, 2018).

Similar conclusions with respect to adverse consequences for health are drawn with regard to social aspects of people’s lives, in particular to social isolation which has been shown to affect both mental and physical health (Miyawaki, 2015). A socially inclusive approach toward older people is related to lower suicide mortality (Yur’yev et al., 2010), while Myck et al. (2015) provide some evidence that social deprivation, broadly conceptualized as deprivation from valuable social interaction, is correlated with life satisfaction. Other studies document that social isolation is related to increased risk of cardiovascular diseases (Barth, Schneider, & von Känel, 2010), infections (Cohen, Doyle, Skoner, Rabin, & Gwaltney, 1997), elevated blood pressure, and deterioration of cognitive abilities (Bassuk, Glass, & Berkman, 1999). A lack of social interactions leads not only to worse health, but consequently also to increased mortality rate (Shor & Roelfs, 2015). A meta-analysis of 148 studies relating social relationships and mortality revealed that overall stronger social relationships lead to a 50% increase in survival likelihood (Holt-Lunstad, Smith, & Layton, 2010).

Obtaining Unbiased Estimates of Impact of Deprivation on Health

In our analysis, we attempt to identify the relationship between health and social and material deprivation. This could be summarized in the following equation:

h_{i, t} = β_{1}^{'} X_{i, t} + γ^{'} d_{i, t} + ε_{i, t},

(1)

where h_i,_t is the health of individual i at time t; X_i,_t are his or her characteristics, d_i,_t is the individual-level indicator (or a vector of indicators) of deprivation, that is, our key variable of interest; and ɛ_i,t is an individual, t-specific residual. Because under this specification there may be factors in ɛ_i,t which are correlated with both d_i,_t and health, the estimated coefficient on deprivation is likely to be biased. An example of such an omitted variable is “ability,” which can clearly affect both deprivation and health and for which we are unlikely to control fully through X_i,_t To reduce such bias, we propose to estimate the following equation, where deprivation and other characteristics are regressed not on levels but rather on changes in health:

Δ h_{i, t} = β_{1}^{'} X_{i, t - 1} + γ^{'} d_{i, t - 1} + ϵ_{i, t}

(2)

with ∆ $h_{i, t}$ defined as (h_i,_t − h_i,_t−1) and the change being conditional on characteristics observed at time t – 1. While this approach does not completely do away with the potential for the omitted variable bias (e.g., ability can affect both deprivation and the rate of health deterioration), it is generally the case that the degree of such bias is going to be reduced compared with Equation 1. The remaining bias may still relate to both fixed and variable factors which are presented in Equation 3 as components of ϵ_i,t:

ϵ_{i, t} = μ_{i} + v_{i, t} .

(3)

Ability could be treated as such a fixed effect, but µ_i can include other characteristics. To account for such a possibility in our approach—given the richness of the SHARE data—we control for a set of such fixed characteristics by including important information on the family socioeconomic situation of the respondent at the age of 10 and his or her health status during childhood (Z_i). A detailed description of childhood variables included in the analysis may be found in Online Appendix C. Thus, the second specification we estimate takes the following form:

Δ h_{i, t} = β_{1}^{'} X_{i, t - 1} + β_{2}^{'} Z_{i} + γ^{'} d_{i, t - 1} + μ_{i} + v_{i, t} .

(4)

In addition, to further control for the potential correlation between health and material and social conditions at time t – 1, we extend this specification to include a vector of health measures (other than h_i,_t) as observed at t – 1, K_i,_t−1:

Δ h_{i, t} = β_{1}^{'} X_{i, t - 1} + β_{2}^{'} Z_{i} + γ^{'} d_{i, t - 1} + β_{3}^{'} K_{i, t - 1} + μ_{i} + v_{i, t} .

(5)

As we shall see in our results, this extension in many instances proves very important in reducing the magnitude of the γ coefficients, thus confirming the strong relationship between health and social and material conditions at any given point.

Through the extensive set of controls and because we estimate the specification in changes of health and not in levels, we argue that we minimize the potential bias on the γ coefficient (or vector of coefficients) of interest, putting the relationship between health and material and social conditions to a very demanding test.

In the final model, we examine the probability of changing the health status from good to bad using the probit regression approach (conditional on good health status in t – 1) for five health outcomes described further in “Data and Method” section. The estimated model—in the full specification—thus takes the following form:

P (H_{i, t} = 1 | H_{i, t - 1} = 0) = β_{1}^{'} X_{i, t - 1} + β_{2}^{'} Z_{i} + β_{3}^{'} K_{i, t - 1} + γ^{'} d_{i, t - 1} + μ_{i} + v_{i, t},

(5)

where H_i,_t is a binary health indicator taking value 1 in the case of bad health.

Under the specifications outlined in Equations 2, 4, and 5, we relate changes from good to bad health status (H_i,t) between Waves 5 and 6 of SHARE to general individual or household characteristics as observed in Wave 5, X_i,_t−1, and a vector of measures of deprivation as recorded in Wave 5, d_i,_t−1 (Specification 1). The X_i,_t−1 characteristics include gender, age polynomial, level of education, living with a partner as well as country dummies, and controls for the time between interview in the two waves in case of all outcomes but death. This set of conditioning variables is then extended to include childhood fixed effects, Z_i, in Specification 2 and, in addition to that, health status, K_i,_t−1, using variables other than H_i,_t in Specification 3. The full list of childhood (Z_i) and health controls (K_i,t₋₁) is available in Online Appendix C. Standard errors are clustered at the country level. We estimate and report results as marginal effects.

Data and Method

Measuring Material and Social Deprivation in SHARE

The SHARE is a multidimensional study conducted in European countries and Israel that concentrates on living conditions of individuals aged 50+, including health, employment status, material situation, and social relations. Since 2004, seven waves of data collection covering almost 300,000 interviews have been conducted. In this article, we focus on information on current aspects of life drawn from Waves 5 and 6 of the SHARE study, which were administered in years 2013 and 2015, respectively. This information is supplemented with additional variables on childhood conditions recovered from the retrospective Wave 3 of SHARE conducted in 2009. For individuals who had not participated in SHARE at this time, information on childhood conditions was collected in Wave 5.

Importantly from the point of this analysis, Wave 5 of SHARE contained an extended list of variables specifically gathered to capture broad aspects of well-being in later life and covering material conditions, quality of the neighborhood, and accessibility to services. These items were used to construct social and material deprivation indices for all SHARE respondents in Wave 5 (for more details, see Table 1 and also Adena, Myck, & Oczkowska, 2015; Bertoni, Cavapozzi, Celidoni, & Trevisan, 2015; Myck et al., 2015; Stoeckel & Litwin, 2015). The indices aggregate binary indicators on whether a person is deprived of a specific item using the so-called “hedonic weights” from a multiple regression of all items on a single life-satisfaction measure (for more details on hedonic weighting scheme, see, for example, Bertoni et al., 2015). The final deprivation indices take values from 0 to 1, with higher values implying higher levels of deprivation. On top of the measures of deprivation in material and social dimension, an index of severe deprivation was introduced to serve as indicator of risk of social exclusion (Myck et al., 2015). Individuals are identified as socially excluded if they are allocated above the 75th percentile of the total distribution of each dimension of deprivation (see also Figures 4a and 4b in Online Appendix B). For the purpose of common reference and cross-country comparison, the threshold is universal for all SHARE countries.

Table 1.

Items Used to Build Material and Social Deprivation Indices in SHARE Wave 5.

Material deprivation index	Social deprivation index
Affordability of	Less than one room per household member
Meat/fish/chicken	Poor reading or writing skills
Fruits/vegetables	Poor computer skills
An unexpected expense	Not feeling part of the local area
A week-long vacation once a year	Vandalism is a problem in the local area
Regular grocery shopping	Local area is not clean
Keeping living costs down by	No helpful people in the local area
Reducing heating	Having difficulties to access:
Wearing worn-out shoes	Bank
Wearing worn-out clothes	Grocery shop
Not replacing glasses	Pharmacy
Postponing dentist	GP
Postponing doctor	Waiting too long to see a doctor
Having payment arrears	Not attending any course or club
Lack of money prevents oneself from doing things that she would like to do	Not taking part in any organization
	Not trusting people
	Feeling left out of things

Source. Bertoni, Cavapozzi, Celidoni, and Trevisan (2015); Myck, Najsztub, and Oczkowska (2015).

Note. SHARE = Survey of Health, Ageing and Retirement in Europe; GP = general practitioner.

While the entire distribution of the material and social indices for each country participating in Wave 5 of SHARE may be found in Figures A1 to A2 in Online Appendix B, in Figure 1 we show the country-specific average levels of deprivation in each domain with relatively high variation across countries and domains. The lowest average level of both material and social deprivation may be seen among the Danish population (0.05 and 0.10, respectively). Estonia clearly stands out as a country with the highest level of material deprivation (0.32), although at the same time the social dimension of deprivation in Estonia is not much different from countries like Israel or Spain. Italy has the highest level of social deprivation (0.25) with the second highest level in the material dimension (0.23). Interestingly, with the exceptions of Estonia, Slovenia, and Spain, the average country level of social deprivation is higher than the material one, and in some cases, these differences are substantial. Figure 2 presents the percentage of the population at risk of social exclusion as defined for the purpose of this analysis. In line with the results captured for the separate indices of deprivation, while as many as 28% of Estonians, 25% of Italians, and 21% of Israelis are at risk of social exclusion, only 2% of the Danish population are identified as being at risk.

Figure 1.

Average levels of MD and SD indices in SHARE Wave 5 by country.

Figure 2.

Proportion of population at risk of social exclusion in SHARE Wave 5 by country.

Health Changes Between Wave 5 and Wave 6 of SHARE

Data in Wave 5 of SHARE were collected in 2013 in 15 countries, of which all but one also took part in the following wave of the survey 2 years later. In Table 2, we present the key descriptive sample statistics for individuals who participated in Wave 5, conditional on the status of their participation in Wave 6. Out of slightly over 65,000 individuals who participated in SHARE in 2013, 72% completed the survey in Wave 6, while additionally there is information on 3% of individuals who died between Waves 5 and 6. This information was collected in the form of the so-called “end-of-life interviews” conducted with close relatives of deceased respondents.

Table 2.

Sample Statistics of Participants of SHARE Wave 5.

	Participated in Wave 6	Deceased	Did not participate in Wave 6
Number of observations	46,744	1,944	16,352
% female	56.2	45.1	54.2
Average age in Wave 5 (years)	66.5	77.9	65.9
Number of observations by country
Austria	3,045	88	1,159
Germany	4,204	93	1,287
Sweden	3,512	93	910
The Netherlands	0	0	4,118
Spain	5,063	360	1,156
Italy	3,596	144	920
France	3,117	82	1,225
Denmark	3,295	116	645
Switzerland	2,546	55	388
Belgium	4,273	149	1,110
Israel	1,768	118	675
Czech Republic	4,337	249	951
Luxembourg	1,089	18	484
Slovenia	2,366	92	456
Estonia	4,533	287	868

Source. SHARE Wave 5 and 6 data, release 6.1.1.

Note. SHARE = Survey of Health, Ageing and Retirement in Europe.

In our main analysis, we use the subsample of Wave 5 participants who completed the interview in Wave 6 or for whom we have information from the end-of-life interview conducted in Wave 6, which is used to identify mortality. We use a large set of variables, and thus, the final analytical sample is limited to individuals for whom all necessary information is available. From 36,644 such individuals, 35,629 completed the interview in Wave 6, and for the further 1,015, we have information from the end-of-life interview (see Tables 3 and 4).

Table 3.

Descriptive Statistics: Participants of Both Waves 5 and 6 of SHARE.

	Wave 5
	Mean		SD
Female	0.534		0.499
Age (years)	65.012		10.033
Education
Primary or none	0.237		0.425
Secondary	0.541		0.498
Tertiary	0.222		0.416
Partner in household	0.675		0.467
Deprivation and exclusion
Material deprivation	0.144		0.195
Social deprivation	0.174		0.136
Social exclusion	0.118		0.323
Number of observations	35,629
	Wave 5		Wave 6
	Mean	SD	Mean	SD
Proportion in bad health
1+ ADL	0.086	0.280	0.098	0.298
1+ MOBILITY	0.447	0.497	0.452	0.498
Poor SAH	0.081	0.272	0.080	0.271
4+ EURO-D	0.261	0.439	0.257	0.437

Source. SHARE Wave 5 and 6 data, release 6.1.1.

Note. Weighted using individual weights. SHARE = Survey of Health, Ageing and Retirement in Europe; ADL = activities of daily living; MOBILITY = difficulties in mobility; SAH = self-reported overall health; EURO-D = depression symptoms as measured by the EURO-D scale.

Table 4.

Descriptive Statistics: Participants of Wave 5 Who Had End-of-Life Interview in Wave 6 of SHARE.

	Wave 5
	M	SD
Female	0.436	0.496
Age (years)	76.104	10.356
Education
Primary or none	0.409	0.492
Secondary	0.476	0.500
Tertiary	0.115	0.319
Partner in household	0.564	0.496
Deprivation and exclusion
Material deprivation	0.169	0.196
Social deprivation	0.273	0.159
Social exclusion	0.237	0.425
Proportion in bad health:
1+ ADL	0.373	0.484
1+ MOBILITY	0.751	0.432
Poor SAH	0.335	0.472
4+ EURO-D	0.480	0.500
Number of observations.	1,015

Source. SHARE Wave 5 and 6 data, release 6.1.1.

Our principal focus in this article is the relationship between the level of material and social deprivation and changes in health status between the two waves. For this purpose, we select a number of key health measures from those available in the SHARE survey and complement the analysis with mortality information from the end-of-life interviews. The following measures of health are used in the analysis:

Limitations with activities of daily living (ADL),

Difficulties in mobility (MOBILITY),

Self-reported overall health status (SAH),

Depression symptoms as measured by the EURO-D scale (EURO-D),

Mortality.

While the indicator of ADL captures problems reported by the respondent with six basic everyday life activities (like dressing or bathing), mobility impairments such as walking 100 m or climbing one flight of stairs are reflected in the MOBILITY Scale (full list of limitations with ADL and MOBILITY can be found in Online Appendix A). For the purpose of further analysis, we assume that an individual is in bad health when at least one of the respective limitations is reported (Chan, Kasper, Brandt, & Pezzin, 2012; Fernandes, Meijer, & Zamarro, 2008). In SHARE, individuals assess their overall health status (SAH) on a 5-point scale from excellent through very good, good, fair, to poor. Here, poor self-assessed health is taken to reflect bad health. Depression symptoms are measured using the EURO-D scale of 12 items capturing, for instance, whether a person had any hopes for the future or had trouble sleeping recently (full list of symptoms in Online Appendix A). The usual cutoff point used in the literature indicating the risk of depression is set at four symptoms (Castro-Costa et al., 2007; Prince et al., 1999).

Descriptive statistics of health outcomes are presented in Table 3 for the sample of respondents who participated in both Waves 5 and 6 of SHARE and in Table 4 for the sample of Wave 5 participants for whom end-of-life interviews were conducted in Wave 6. For participants of both Waves 5 and 6, we observe a natural deterioration in physical health during the 2-year period, but in terms of mental health we observe a slight improvement.

The measures of health status in our data are strongly differentiated by country, both in terms of the level of health in a specific wave and in terms of changes between waves. This is reflected in Figure 3a to 3d where we show the proportions of respondents in bad health among those who were interviewed in both waves.

Figure 3.

Proportion of respondents in bad health in Waves 5 and 6 of SHARE by country: (a) proportion 1+ ADL, (b) proportion 1+ MOBILITY, (c) proportion in poor health (SAH), and (d) proportion 4+ EURO-D.

Results

In Tables 5 and 6 we present results of the estimations in the three specifications described in “Obtaining Unbiased Estimates of Impact of Deprivation on Health” section of the article for each of the four measures of health from the “Health Changes Between Wave 5 and Wave 6 of SHARE” section, while in Table 7 we show results of estimations for the probability of death between Waves 5 and 6. In Table 5, we show results when controlling for material and social deprivation indices, whereas in Table 6 the two are combined in a single social exclusion indicator. All results are reported as marginal effects. Given a high number of conditioning variables used in the analysis, only a subset of results is presented in the tables.

Table 5.

Role of Material and Social Deprivation for the Probability of Changes in Health Between SHARE Wave 5 and Wave 6 (Marginal Effects).

	(1)	(2)	(4)	(5)	(6)	(8)	(9)	(10)	(12)	(13)	(14)	(16)
	1+ ADL			1+ MOBILITY			Poor SAH			4+ EURO-D
	Specification 1	Specification 2	Specification 3	Specification 1	Specification 2	Specification 3	Specification 1	Specification 2	Specification 3	Specification 1	Specification 2	Specification 3
Material deprivation	0.077***(7.89)	0.074***(7.50)	0.032***(4.12)	0.100**(2.75)	0.098**(2.76)	0.058(1.89)	0.073***(10.93)	0.070***(10.53)	0.043***(6.70)	0.148***(8.89)	0.139***(8.41)	0.104***(6.41)
Social deprivation	0.113***(8.65)	0.108***(8.47)	0.041***(4.38)	0.272***(6.34)	0.270***(6.14)	0.190***(4.92)	0.101***(7.94)	0.098***(7.99)	0.047***(4.55)	0.254***(8.02)	0.244***(8.08)	0.184***(6.49)
Female	−0.002(–0.49)	−0.001(–0.38)	−0.025***(–5.39)	0.070***(12.44)	0.070***(12.31)	0.057***(5.05)	−0.005(–1.31)	−0.005(–1.26)	−0.029***(–4.36)	0.070***(16.37)	0.071***(17.78)	0.040***(5.67)
Education (ref. cat.: primary or none)
Secondary	−0.008(–1.72)	−0.005(–1.15)	−0.002(–0.40)	−0.029**(–2.69)	−0.028*(–2.42)	−0.020(–1.89)	−0.012**(–2.62)	−0.009*(–2.23)	−0.009*(–2.25)	−0.013(–1.09)	−0.007(–0.58)	−0.007(–0.56)
Tertiary	−0.025***(–4.84)	−0.020***(–3.57)	−0.012*(–2.19)	−0.067***(–4.80)	−0.066***(–4.30)	−0.053***(–3.68)	−0.021***(–3.68)	−0.017**(–3.27)	−0.014*(–2.54)	−0.018(–1.45)	−0.009(–0.66)	−0.005(–0.42)
Age	0.003***(13.33)	0.003***(13.18)	0.002***(8.83)	0.009***(12.87)	0.009***(12.55)	0.007***(10.24)	0.002***(10.21)	0.002***(9.91)	0.001***(4.22)	0.002***(3.66)	0.002***(3.67)	0.000(0.11)
No partner in household	0.006*(2.32)	0.006*(2.19)	0.006*(2.14)	0.011(1.22)	0.011(1.21)	0.009(1.05)	0.004(1.41)	0.004(1.39)	0.004(1.26)	−0.013*(–2.15)	−0.013*(–2.18)	−0.014*(–2.30)
Childhood controls		incl.	incl.		incl.	incl.		incl.	incl.		incl.	incl.
(t – 1) health controls			incl.			incl.			incl.			incl.
No. of observations	32,690	32,690	32,690	19,611	19,611	19,611	32,925	32,925	32,925	27,181	27,181	27,181

Source. Own calculations using SHARE Waves 5 and 6 data (release 6.1.1).

Note. t statistics in parentheses, standard errors clustered at the country level. In Specification 1, we include controls for demographic information; childhood fixed effects are added in Specification 2; and health controls in time t – 1 are included in Specification 3. Not reported in the table but included in all specifications are the following controls: country dummies, time between waves. For more details on childhood controls and (t – 1) health controls, see Online Appendix C. SHARE = Survey of Health, Ageing and Retirement in Europe; ADL = activities of daily living; MOBILITY = difficulties in mobility; SAH = self-reported overall health; EURO-D = depression symptoms as measured by the EURO-D scale.

p < .05. **p < .01. ***p < .001

Table 6.

Role of Social Exclusion for the Probability of Changes in Health Between SHARE Wave 5 and Wave 6 (Marginal Effects).

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)
	1+ ADL			1+ MOBILITY			Poor SAH			4+ EURO-D
	Specification 1	Specification 2	Specification 3	Specification 1	Specification 2	Specification 3	Specification 1	Specification 2	Specification 3	Specification 1	Specification 2	Specification 3
Social exclusion	0.039***(8.38)	0.037***(7.63)	0.008(1.76)	0.094***(4.92)	0.091***(4.77)	0.053**(3.08)	0.040***(8.08)	0.038***(7.65)	0.019***(4.37)	0.096***(14.63)	0.091***(13.76)	0.065***(8.76)
Female	−0.001(–0.23)	−0.000(–0.09)	−0.027***(–5.60)	0.071***(11.75)	0.071***(11.73)	0.054***(4.52)	−0.004(–0.98)	−0.003(–0.91)	−0.032***(–4.38)	0.070***(15.22)	0.071***(16.63)	0.035***(4.63)
Education (ref. cat.: primary or none)
Secondary	−0.014**(–2.82)	−0.010*(–2.05)	−0.003(–0.85)	−0.039***(–3.51)	−0.035**(–3.04)	−0.024*(–2.32)	−0.018***(–3.58)	−0.013**(–3.04)	−0.011**(–2.75)	−0.024(–1.91)	−0.014(–1.18)	−0.011(–1.00)
Tertiary	−0.036***(–6.60)	−0.029***(–5.02)	−0.015**(–2.83)	−0.085***(–6.31)	−0.079***(–5.36)	−0.061***(–4.42)	−0.032***(–5.03)	−0.025***(–4.51)	−0.018**(–3.18)	−0.040**(–3.08)	−0.024(–1.86)	−0.016(–1.28)
Age	0.003***(14.01)	0.003***(13.82)	0.002***(8.90)	0.009***(12.83)	0.009***(12.49)	0.007***(10.16)	0.002***(10.95)	0.002***(10.57)	0.001***(3.76)	0.002***(4.30)	0.002***(4.25)	−0.000(–0.01)
No partner in household	0.011***(4.01)	0.010***(3.75)	0.008**(2.78)	0.016(1.89)	0.015(1.85)	0.011(1.43)	0.008**(2.77)	0.008**(2.68)	0.005*(2.02)	−0.007(–1.12)	−0.007(–1.20)	−0.010(–1.62)
Childhood controls		incl.	incl.		incl.	incl.		incl.	incl.		incl.	incl.
(t – 1) health controls			incl.			incl.			incl.			incl.
No. of observations	32,690	32,690	32,690	19,611	19,611	19,611	32,925	32,925	32,925	27,181	27,181	27,181

Source. Own calculations using SHARE Waves 5 and 6 data (release 6.1.1).3.

Note. SHARE = Survey of Health, Ageing and Retirement in Europe; ADL = activities of daily living; MOBILITY = difficulties in mobility; SAH = self-reported overall health; EURO-D = depression symptoms as measured by the EURO-D scale.

p < .05. **p < .01. ***p < .001.

Table 7.

Role of Social and Material Deprivation and Social Exclusion for the Probability of Death Between SHARE Wave 5 and Wave 6 (Marginal Effects).

	(1)	(2)	(3)	(4)	(5)	(6)
	Specification 1	Specification 2	Specification 3	Specification 1	Specification 2	Specification 3
Material deprivation	0.014***(3.37)	0.014**(3.20)	−0.003(–0.70)
Social deprivation	0.059***(7.44)	0.059***(7.27)	0.019**(2.66)
Social exclusion				0.016***(7.60)	0.016***(7.08)	0.003(1.24)
Female	−0.019***(–10.75)	−0.019***(–10.50)	−0.035***(–10.66)	−0.019***(–10.03)	−0.019***(–9.71)	−0.035***(–10.71)
No partner in household	0.001(0.56)	0.001(0.52)	0.002(0.92)	0.002(1.02)	0.002(0.95)	0.002(1.00)
Education (ref. cat.: primary or none)
Secondary	−0.004(–0.97)	−0.004(–0.84)	−0.003(–0.74)	−0.006(–1.36)	−0.005(–1.09)	−0.003(–0.76)
Tertiary	−0.008*(–2.07)	−0.008(–1.93)	−0.007(–1.68)	−0.011**(–2.99)	−0.011*(–2.51)	−0.007(–1.77)
Age of respondent at the time of interview	0.002***(20.95)	0.002***(21.26)	0.002***(13.37)	0.003***(22.33)	0.003***(22.87)	0.002***(14.74)
Childhood controls		incl.	incl.		incl.	incl.
(t – 1) health controls			incl.			incl.
No. of observations	38,019	38,019	38,019	38,019	38,019	38,019

Source. Own calculations using SHARE Waves 5 and 6 data (release 6.1.1).

Note. t statistics in parentheses, standard errors clustered at the country level. In Specification 1, we include controls for demographic information; childhood fixed effects are added in Specification 2; and health controls in time t – 1 are included in Specification 3. Not reported in the table but included in all specifications are the following controls: country dummies, time between waves. Time between waves is not included as a control. SHARE = Survey of Health, Ageing and Retirement in Europe.

p < .05. **p < .01. ***p < .001.

In all of the estimated specifications in Table 5, the effects of both deprivation measures are positive and statistically significant, implying strong correlation between material and social deprivation and deterioration of health. For all four health measures, we find only a minimal biasing role of the childhood variables added in Specification 2. The same cannot be said for the comparison with Specification 3, in which the other health measures from Wave 5 are added to the set of control variables. In the case of all health changes, controlling for other measures of health status in time (t – 1) has a strong influence on the value of the estimated effects. For example, for 1+ ADL, adding the (t – 1) health controls in Specification 3 reduces the value of the marginal effect of material deprivation from 0.074 to 0.032 and the value of the effect on social deprivation from 0.108 to 0.041. Slightly smaller (relative) changes are recorded for other health measures. These findings confirm the strong correlation between health level and both material and social deprivation in the baseline period (t – 1) and with changes in the analyzed health dimension.

As for other results in Table 5, it is worth noting that women have a higher probability of falling into depression. Both education (–) and age (+) have the expected signs with regard to the worsening of physical health; however, the effect on mental health is less straightforward.

With regard to the estimation which uses only the indicator of risk of social exclusion (Table 6), it is worth noting that the results reported for the separate indices generally hold, and we also observe the pattern of an adverse role of social exclusion toward physical and mental health changes. The only exception is in Specification 3 for 1+ ADL where we not only find a substantial bias from excluding the additional health indicators from (t – 1) but also—as a result of including these in Specification 3—find that the social exclusion index has no effect on change in this dimension of health.

Finally, looking at the results in the model for the probability of dying between waves (Table 7), we see a similar pattern with respect to the role of controls for (t – 1) health status in Specification 3. What is interesting is the fact that in Specification 3 while material deprivation seems to bear no effect on the probability of dying between the waves, social deprivation continues to play a role, even under the most extended specification. In addition, we find a familiar result that women are less likely to die than men, and as one could expect, the effect is stronger once we condition for health status in Specification 3.

The estimated coefficients on deprivation and exclusion measures presented in Tables 5 to 7 show the average relationships for all countries in the sample. However, as we show in Tables 9 to 11 in Online Appendix D, these average values are very close in magnitude to the effects estimated for four countries for which the samples are the largest (Czech Republic, Germany, Estonia, and Spain), which is reassuring given the institutional and cultural heterogeneity of these four countries.

Given the results presented above—independently of material and social deprivation—one general conclusion is that in the analysis of influence of deprivation on health, it seems more informative to include the two separate indices of deprivation, material and social, rather than the combined indicator for social exclusion. Our results suggests that the two dimensions of deprivation might affect people differently with respect to different measures of health, and this heterogeneity in effects is lost if we use a combined index of exclusion.

Conclusion

Uncovering the causal components in the relationship between health and material and social conditions seems to be one of the more challenging problems in microeconomic and epidemiological studies. The outcomes are so tightly intertwined and so strongly result from common factors such as innate qualities, childhood conditions, and—in fact—the entire life history, that providing a clean causal identification is extremely difficult, if at all possible. This is true for the causal effect of health on material/social conditions and for the effect in the other direction. While this article, in which we address the latter problem, does not provide the definitive answers as to if and to what extent material and/or social conditions influence health, for a number of reasons we believe it makes an important contribution to the literature on the subject.

In our analysis, which is based on the data from Waves 5 and 6 of the SHARE, we examine the implications of several factors for health among the 50+ European population, with the focus on complex and comprehensive measures of material and social conditions. We take a very careful approach to the examined specifications to judge the potential extent of the endogeneity bias and to minimize it in the final approach.

To do that, first, instead of taking levels we examine the relationship with respect to changes in a number of dimensions of health status between two points in time. In particular, the analysis refers to three dimensions of physical health, a measure of mental health, and the likelihood of dying. In the basic approach, in which we develop the methodology, this covers the changes between Waves 5 and 6 of the SHARE survey. Second, to control for the factors which may be considered as “initial conditions” jointly determining late life health and socioeconomic outcomes, we extend our basic specification to include a long range of variables relating to childhood conditions of SHARE respondents. Finally, in the last specification, in the examination of changes in each of the five definitions of health, we control for the levels in additional health dimensions at time (t – 1) to account for the correlation of material and social conditions with health status at the initial point in time.

As we show in our results for all physical measures of health, the last extension of the list of control variables matters most in terms of its influence on the estimated coefficients of interest. Still, even though the magnitude of the coefficients falls by about half in comparison with the basic specification, the estimates suggest a strong and statistically significant relationship between the measures of deprivation and social exclusion and health. Interestingly, in our examination of the probability of dying, we find that this is only affected by social and not by material deprivation.

One important limitation of the analysis is its heavy reliance on detailed specific questions with which we identify the degree of deprivation. This battery of questions was specifically designed for the SHARE survey, but as a full set the questions were only asked in Wave 5 of the survey. The approach is therefore not easily transferable to other policy and cultural contexts. We believe, though, that this and other analyses which were based on these deprivation indicators stress the importance of collecting such information and continued development of deprivation measures in the material and social dimensions.

As we argued above, our approach to the analysis of the relationship between health and material and social conditions should allow for causal interpretation of the results in our final specification. The article thus confirms earlier findings that material and social conditions in old age have their causal implications for the development of health status. Our results extend beyond the existing research and provide evidence that the impact on changes in health should not only be regarded separately with respect to the two dimensions of deprivation but can also be considered when material and social disadvantages are combined. The results seem to be of potentially high relevance from the policy perspective, and examination of the specific role of material and social deprivation for developments in health at a later stage in people’s lives can be instrumental in designing specifically targeted policy interventions. Along these lines, understanding of negative health developments among the currently deprived population may serve as an important sign of future developments in the level of health and health care needs, providing policy makers with some expectations for the future needs and the resulting costs.

Supplemental Material

appendices_reviewed – Supplemental material for Implications of Social and Material Deprivation for Changes in Health of Older People

Supplemental material, appendices_reviewed for Implications of Social and Material Deprivation for Changes in Health of Older People by Michał Myck, Mateusz Najsztub and Monika Oczkowska in Journal of Aging and Health

Footnotes

Acknowledgements

We are grateful for comments and suggestions to participants in the MASEP conference (Łódź 2017), of the SHARE working group meeting (Sofia, 2017) and of several meetings organised within the ROSEnet COST Action (CA 15122). We also thank two anonymous referees for their suggestions and comments, which helped to considerably improve the paper, and Kajetan Trzciński for careful proof-reading. This article uses data from SHARE Waves 3, 5, and 6 (DOIs: 10.6103/SHARE.w3.600, 10.6103/SHARE.w5.600, 10.6103/SHARE.w6.600; [1-3]); see [4] for methodological details. The SHARE data collection has been primarily funded by the European Commission through FP5 (QLK6-CT-2001-00360), FP6 (SHARE-I3: RII-CT-2006-062193, COMPARE: CIT5-CT-2005-028857, SHARELIFE: CIT4-CT-2006-028812), and FP7 (SHARE-PREP: N°211909, SHARE-LEAP: N°227822, SHARE M4: N°261982). Additional funding from the German Ministry of Education and Research, the Max Planck Society for the Advancement of Science, and the U.S. National Institute on Aging (U01_AG09740-13S2, P01_AG005842, P01_AG08291, P30_AG12815, R21_AG025169, Y1-AG-4553-01, IAG_BSR06-11, OGHA_04-064, HHSN271201300071C) and from various national funding sources is gratefully acknowledged (see ).

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Initial analysis and conceptual development of the results presented in this article were funded through the project “Support of realization of panel surveys of individuals aged 50+ in the international project Survey of Health, Ageing and Retirement in Europe (SHARE)” co-financed from the European Social Fund and conducted in cooperation with the Polish Ministry of Family, Labour and Social Policy. Views and opinions expressed in this publication are those of the authors and do not necessarily reflect the official position of the Polish Ministry of Family, Labour and Social Policy or the Centre for Economic Analysis. The usual disclaimer applies.

ORCID iD

Monika Oczkowska

Supplemental Material

Supplemental material for this article is available online.

References

Adena

Myck

(2014). Poverty and transitions in health in later life. Social Science & Medicine, 116, 202-210. doi:10.1016/j.socscimed.2014.06.045

Adena

Myck

Oczkowska

(2015). Material deprivation items in SHARE wave 5 data: A contribution to a better understanding of differences in material conditions in later life. In Börsch-Supan

Kneip

Litwin

Myck

Weber

(Eds.), Ageing in Europe—Supporting policies for an inclusive society (pp. 25-38). Berlin, Germany: De Gruyter. doi:10.1515/9783110444414-004

Atkinson

A. B.

Hills

(1998). Exclusion, employment and opportunity [Monograph]. Retrieved from http://sticerd.lse.ac.uk/case

Barth

Schneider

von Känel

(2010). Lack of social support in the etiology and the prognosis of coronary heart disease: A systematic review and meta-analysis. Psychosomatic Medicine, 72, 229-238. doi:10.1097/PSY.0b013e3181d01611

Bassuk

S. S.

Glass

T. A.

Berkman

L. F.

(1999). Social disengagement and incident cognitive decline in community-dwelling elderly persons. Annals of Internal Medicine, 131, 165-173.

Benach

Yasui

Borrell

Sáez

Pasarin

(2001). Material deprivation and leading causes of death by gender: Evidence from a nationwide small area study. Journal of Epidemiology & Community Health, 55, 239-245. doi:10.1136/jech.55.4.239

Bertoni

Cavapozzi

Celidoni

Trevisan

(2015). Development and validation of a material deprivation index. In Börsch-Supan

Kneip

Litwin

Myck

Weber

(Eds.), Ageing in Europe—Supporting policies for an inclusive society (pp. 57-66). Berlin, Germany: De Gruyter. doi:10.1515/9783110444414-007

Bhalla

Lapeyre

(1997). Social exclusion: Towards an analytical and operational framework. Development and Change, 28, 413-433. doi:10.1111/1467-7660.00049

Cambois

Solé-Auró

Robine

J.-M.

(2016). Economic hardship and educational differentials in disability in 26 European countries. Journal of Aging and Health, 28, 1214-1238. doi:10.1177/0898264316656503

10.

Carstairs

Morris

(1989). Deprivation: Explaining differences in mortality between Scotland and England and Wales. British Medical Journal, 299,886-889.

11.

Castro-Costa

Dewey

Stewart

Banerjee

Huppert

Mendonca-Lima

. . . Prince

(2007). Prevalence of depressive symptoms and syndromes in later life in ten European countries: The SHARE study. The British Journal of Psychiatry, 191, 393-401.

12.

Chakravarty

S. R.

D’Ambrosio

(2006). The measurement of social exclusion. Review of Income and Wealth, 52, 377-398. doi:10.1111/j.1475-4991.2006.00195.x

13.

Chan

K. S.

Kasper

J. D.

Brandt

Pezzin

L. E.

(2012). Measurement equivalence in ADL and IADL difficulty across international surveys of aging: Findings from the HRS, SHARE, and ELSA. The Journals of Gerontology, Series B: Psychological Sciences & Social Sciences, 67, 121-132.

14.

Cohen

Doyle

W. J.

Skoner

D. P.

Rabin

B. S.

Gwaltney

J. M.

(1997). Social ties and susceptibility to the common cold. Journal of the American Medical Association, 277, 1940-1944. doi:10.1001/jama.1997.03540480040036

15.

Deaton

(2008). Income, health, and well-being around the world: Evidence from the Gallup World Poll. Journal of Economic Perspectives, 22(2), 53-72. doi:10.1257/jep.22.2.53

16.

Detollenaere

Desmarest

A.-S.

Boeckxstaens

Willems

(2018). The link between income inequality and health in Europe, adding strength dimensions of primary care to the equation. Social Science & Medicine, 201, 103-110. doi:10.1016/j.socscimed.2018.01.041

17.

Doebler

Glasgow

(2017). Relationships between deprivation and the self-reported health of older people in Northern Ireland. Journal of Aging and Health, 29, 594-619. doi:10.1177/0898264316641079

18.

Ettner

S. L.

(1996). New evidence on the relationship between income and health. Journal of Health Economics, 15, 67-85. doi:10.1016/0167-6296(95)00032-1

19.

Fernandes

Meijer

Zamarro

(2008). Comparison between SHARE, ELSA, and HRS. In Börsch-Supan

, et al. (Eds.). Mannheim, Germany: Mannheim Research Institute for the Economics of Aging. Available from http://www.share-project.org

20.

Gaertner

(1993). Amartya Sen: Capability and well-being. Oxford University Press. Retrieved from http://www.oxfordscholarship.com/view/10.1093/0198287976.001.0001/acprof-9780198287971-chapter-5

21.

Gero

Kondo

Shirai

Kawachi

(2017). Associations of relative deprivation and income rank with depressive symptoms among older adults in Japan. Social Science & Medicine, 189, 138-144. doi:10.1016/j.socscimed.2017.07.028

22.

Gunasekara

Carter

Crampton

Blakely

(2013). Income and individual deprivation as predictors of health over time. International Journal of Public Health, 58, 501-511.

23.

Haan

Myck

(2009). Dynamics of health and labor market risks. Journal of Health Economics, 28, 1116-1125. doi:10.1016/j.jhealeco.2009.09.001

24.

Holt-Lunstad

Smith

T. B.

Layton

J. B.

(2010). Social relationships and mortality risk: A meta-analytic review. PLoS Medicine, 7(7), e1000316. doi:10.1371/journal.pmed.1000316

25.

Jehoel-Gijsbers

Vrooman

(2008). Social exclusion of the elderly: A comparative study of EU member states.

26.

ENEPRI Research Report No. 57. Available at SSRN: https://ssrn.com/abstract=2027886 or https://dx-doi-org.web.bisu.edu.cn/10.2139/ssrn.2027886.

27.

Levitas

Pantazis

Fahmy

Gordon

Lloyd

Patsios

(2007, January). The multi-dimensional analysis of social exclusion [Monograph]. Retrieved from http://roar.uel.ac.uk/1781/

28.

Marmot

(2002). The influence of income on health: Views of an epidemiologist. Health Affairs, 21(2), 31-46. doi:10.1377/hlthaff.21.2.31

29.

Miyawaki

C. E.

(2015). Association of social isolation and health across different racial and ethnic groups of older Americans. Ageing & Society, 35, 2201-2228. doi:10.1017/S0144686X14000890

30.

Myck

Najsztub

Oczkowska

(2015). Measuring social deprivation and social exclusion. In Börsch-Supan

Kneip

Litwin

Myck

Weber

(Eds.), Ageing in Europe—Supporting policies for an inclusive society (pp. 67-78). Berlin, Germany: De Gruyter. doi:10.1515/9783110444414-008

31.

Nussbaum

M. C.

(2001). Women and human development: The capabilities approach. New York, USA: Cambridge University Press.

32.

Pirani

(2013). Evaluating contemporary social exclusion in Europe: A hierarchical latent class approach. Quality & Quantity, 47, 923-941. doi:10.1007/s11135-011-9574-2

33.

Prince

Reischies

Beekman

Fuhrer

Jonker

Kivela

. . . Copeland

(1999). Development of the EURO-D scale—A European Union initiative to compare symptoms of depression in 14 European centres. The British Journal of Psychiatry, 174, 330-338.

34.

Rohde

Tang

K. K.

Osberg

Rao

D. S. P.

(2017). Is it vulnerability or economic insecurity that matters for health? Journal of Economic Behavior & Organization, 134, 307-319. doi:10.1016/j.jebo.2016.12.010

35.

Sen

(2000). Social exclusion: Concept, application, and scrutiny. Manila, Philippines: Office of Environment and Social Development, Asian Development Bank.

36.

Shor

Roelfs

(2015). Social contact frequency and all-cause mortality: A meta-analysis and meta-regression. Social Science & Medicine, 128, 76-86. doi:10.1016/j.socscimed.2015.01.010

37.

Stoeckel

Litwin

(2015). Accessibility to neighbourhood services and well-being among older Europeans. In Börsch-Supan

Kneip

Litwin

Myck

Weber

(Eds.), Ageing in Europe—Supporting policies for an inclusive society (pp. 39-48). Berlin, Germany: De Gruyter. doi:10.1515/9783110444414-005

38.

Teerawichitchainan

Knodel

(2015). Economic status and old-age health in poverty-stricken Myanmar. Journal of Aging and Health, 27, 1462-1484. doi:10.1177/0898264315584577

39.

Testi

Ivaldi

(2009). Material versus social deprivation and health: A case study of an urban area. The European Journal of Health Economics, 10, 323-328. doi:10.1007/s10198-008-0136-z

40.

Townsend

(1979). Poverty in the United Kingdom: A survey of household resources and standards of living. Berkeley and Los Angeles, USA: University of California Press.

41.

Tubeuf

Jusot

(2011). Social health inequalities among older Europeans: The contribution of social and family background. The European Journal of Health Economics, 12, 61-77. doi:10.1007/s10198-010-0229-3

42.

Walsh

Scharf

Cullinan

Finn

(2012). Deprivation and its measurement in later life: Findings from a mixed-methods study in Ireland. Galway: Irish Centre for Social Gerontology.

43.

Walsh

Scharf

Keating

(2016). Social exclusion of older persons: A scoping review and conceptual framework. European Journal of Ageing, 14,81-98.

44.

Winkleby

M. A.

Jatulis

D. E.

Frank

Fortmann

S. P.

(1992). Socioeconomic status and health: How education, income, and occupation contribute to risk factors for cardiovascular disease. American Journal of Public Health, 82, 816-820. doi:10.2105/AJPH.82.6.816

45.

Yur’yev

Leppik

Tooding

L.-M.

Sisask

Värnik

(2010). Social inclusion affects elderly suicide mortality. International Psychogeriatrics, 22, 1337-1343. doi:10.1017/S1041610210001614

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

0.58 MB