Psychological Well-Being in Italian Families: An Exploratory Approach to the Study of Mental Health Across the Adult Life Span in the Blue Zone

Self-reported measures of psychological well-being and depressive symptoms were examined across differently aged family members, while controlling for the impact of marital status and personal satisfaction about family and non-family relations. Twenty-one grandchildren (i.e., ages 21-36 years) were recruited with their parents (i.e., 48-66 years old) and grandparents (i.e., 75-101 years of age) in the ‘blue zone’ of Ogliastra, an Italian area known for the longevity of its inhabitants. Each participant was individually presented a battery of questionnaires assessing their lifestyle and several perceived mental health indices, including the Warwick-Edinburgh Mental Well-Being Scale (WEMWBS, Tennant et al., 2007), and the Center for Epidemiologic Studies Depression Scale (i.e., CES-D, Radloff, 1977). After assessing the level of concordance among adults sharing the same context, the Hierarchical Linear Modeling (HLM) approach was used to assess the nested dataset. It was found that family membership (i.e., grandchildren versus parents and grandparents) predicted the WEMWBS score but not the CES-D when the impact of marital status and personal satisfaction about social (i.e., family and non-family) ties was controlled for. Moreover, two separate repeated-measure Analyses of Variance (ANOVAs) documented similar level of personal satisfaction about social relationships across the three family groups. In conclusions, satisfying social ties with friends and family members together with an active socially oriented life style seems to contribute to the promotion of mental health in adult span.

relationships. In this regard, Antonucci and Akiyama (1987) proposed the 'convoy model' according to which individuals are immersed in dynamic social networks within which reciprocal exchanges shape an individual's mental health across the entire life span. More recently, Margrett et al. (2011) found that greater perceived social support was associated with various outcomes including increased life satisfaction, greater extroversion and lower loneliness in a sample of American octogenarians and centenarians. Consistent with this, there is empirical evidence that positive family ties provide instrumental support for the maintenance of adequate levels of psychological well-being (e.g., Antonucci, Birditt, Sherman, & Trinh, 2011), especially in mid-to late-life (e.g., Ward, Spitze, & Deane, 2009). However, the impact of number of family ties on mental health of family members is quite controversial. Indeed, Peek and Lin (1999) found fewer depressive signs in people with a higher number of relatives in their social network than in individuals with a lower numbers of family members. In contrast, Haines, Beggs, and Hurlbert (2008) did not find any effect of the family proportion in shaping psychological well-being in adult life span. More recently, Fuller-Iglesias, Webster, and Antonucci (2015) conducted a longitudinal study on almost nine hundred young, middle-aged and elderly family members. The authors found that the size and composition of the family did not influence self-reported psychological wellbeing (i.e., occurrence of depressive signs) in young (i.e., 18-34 years) and middle-aged (i.e., 35-49 years) participants. In contrast, older adults (i.e., 50-80 years) receiving greater support from many family members reported fewer depressive signs over time, whereas young and middle-aged participants reporting a higher negative quality of family relationships (i.e., negative aspects such as conflict and burden) presented more depressive symptoms.
Growing evidence indicates that social support from friends is also relevant for the maintenance of physical and mental health in late adult span (e.g., Antonucci, 2001;Fuller-Iglesias, 2015;Fuller-Iglesias et al., 2015;Poon & Cohen-Mansfield, 2011). More specifically, recent studies have documented that active involvement in social activities provides elderly people with positive relationships that contribute to well-being (e.g., Everard, Lach, Fisher, & Baum, 2000;Litwin & Shiovitz-Ezra, 2006) and perceived physical health (Fuller-Iglesias, 2015). This seems crucial because further evidence highlights that depressive symptoms are highest among the oldest-old (Gostynski et al., 2002), who are most vulnerable to loneliness and a lack of social support (Dykstra, 2009). This is consistent with a wider literature (e.g., Cornwell & Laumann, 2015;Poon & Cohen-Mansfield, 2011) that confirms that social connectedness, even in late adulthood, can facilitate the access to social opportunities, improve self-esteem, increase physical activity and significantly reduce loneliness. One possible scenario is that the support of family members and friends across adult life span lowers psychological stress and depression, enhances cooperation and thereby reduces loneliness (Dykstra, 2009;Stroebe, Zech, Stroebe, & Abakoumkin, 2005).
The foregoing evidence is also consistent with findings from the Italian context, where the current study was conducted. For example, there are indications that more intense social support from family members and neighbors is associated with fewer depressive symptoms in later life (e.g., Carpiniello, Carta, & Rudas, 1989).
In addition, de Belvis et al. (2008) found that fewer social contacts with friends is predictive of a decline in life quality among non-institutionalised Italian older people (i.e., individuals aged 60 years or more). Most recently,  found that young adults and older people living in urban and rural area  Fastame, Penna, Rossetti, & Agus, 2014). Moreover, Fastame, Penna, Rossetti, and Agus (2014) reported that higher selfperceived support from non-family members (i.e., respect and support from the younger generations) was associated to greater general psychological well-being, better self-efficacy and fewer depressive signs in late adulthood. According to , these positive outcomes may be because collectivist principles are highly valued in traditional Sardinian culture. Such speculation is consistent with previous studies (e.g., Martin, Hagberg, & Poon, 1997;Rook & Pietromonaco, 1987) indicating that the number and/or quality of social contacts represent a vital resource for older people.
The living arrangements of traditional Sardinian families typically promote collectivistic values; during late adulthood people are more likely to cohabit with adult children or to receive significant support from them (e.g., de Jong Gierveld & Van Tilburg, 1999;Hank, 2007). Although this suggests that familial networks may maintain the mental health of Sardinian adults, direct empirical support is lacking.
The current study was aimed at exploring the role of family membership (i.e., grandchild versus parent and grandparent) as a predictor of various mental health (i.e., psychological well-being and depressive symptoms) outcomes. Potentially confounding effects of marital status, and self-reported measures of personal satisfaction about family and non-family relationships were statistically controlled for.
An additional aim was to examine whether the mental health outcomes (i.e., psychological well-being, depressive measures), perceived physical health, satisfaction with family, and non-family ties varied across the three generations.
In the present study, psychological well-being was assessed using a recently validated Italian version (Gremigni & Stewart-Brown, 2011) of the Warwick-Edinburgh Mental Well-Being Scale (i.e., WEMWBS; Tennant et al., 2007). Depressive symptoms were determined using the Italian version of the Center for Epidemiologic Studies Depression Scale (i.e., CES-D, Radloff, 1977). Despite the wide use of these tools, to our knowledge, this is the first investigation using them to assess the impact of social networks on mental health. This would also appear to be the first study assessing mental health outcomes among family members recruited exclusively within the Sardinian 'blue zone', a longevity hot-spot (Poulain et al., 2004). In addition, to our knowledge, no previous studies reported the use of the WEMWBS (Tennant et al., 2007) to assess psychological well-being in very old Italian adults (i.e., over 82 years of age).
The lack of prior evidence limits speculation about certain aspects of this study; however, various hypotheses can be stated. First, it was hypothesized that family membership, together with marital status, perceived personal satisfaction about family and friend relationships would predict self-reported psychological well-being and depressive symptoms (e.g., Fastame, Penna, Rossetti, & Agus, 2014;Fuller-Iglesias et al., 2015;Kitayama et al., 2000). Second, mean depression scores were expected to be significantly lower than the Italian cut-off score as previously reported (e.g., Fastame, Penna, Rossetti, & Agus, 2014). Finally, in a collectivistic rural community where the current data were collected (e.g., , similar level of personal satisfaction about family and non-family relationships were expected across the three family generations (i.e., grandchildren versus parents and grandparents) (e.g., Kwan et al., 1997;Kitayama et al., 2000).

Method Participants
Sixty-three cognitively healthy community dwelling adults were recruited in twenty-one families residing in Arzana, a small village (i.e., approximate population 2500) located in the Sardinian 'blue zone' area of Ogliastra (Poulain et al., 2004), which is known for the longevity of its inhabitants and their successful aging (e.g., . Participants were volunteers and were assigned to one of three family status groups, as appropriate: Grandchild, Parent, or Grandparent. Participants were required to meet all of the following inclusion criteria: (a) have been born and currently reside in Arzana; (b) live in a private house, with or without other members of the family; (c) be descendant of people that lived in that Sardinian area for at least two previous generations; (d) be cognitively healthy, that is, having a score ≥ 24/30 on the Mini-Mental State Examination test (MMSE; Folstein, Folstein, & McHugh, 1975); (e) each participant of the grandchildren group had to take part in the study with his/her parent and grandparent, that is, a granddaughter could participate only if her cognitively healthy mother and grandmother accepted to take part in the study (i.e., female lineal descendants), so did the young male adults with their fathers and grandfathers (i.e., male lineal descendants). Although these criteria restricted recruitment they were deemed necessary to isolate the influence of the Sardinian context and to be compatible with related previous research.
Volunteers were recruited through personal contacts of the third author and via direct appeals to local community groups. Respondents did not receive any reward (e.g., financial support) for their participation. Table 1 illustrates the socio-demographic characteristics of the participants. Gender, χ 2 (2, N = 63) = 0.00, p = 1.00, was counterbalanced across the three family groups. Following De Beni et al. (2007), education was dichotomized according to low (i.e., 1-8 years) and high (> 8 years) levels. As one can see in Table 1, education was higher among Grandchildren relative to their Grandparents, χ 2 (2, N = 63) = 38.20, p < .001. This is not surprising. Younger generations from this region now have far greater opportunities to remain in formal education whereas older generations typically entered farming-related occupations following only basic formal education.

Materials
Each participant first read and provided written informed consent to participate and were then presented the following questionnaires/test: The Mini-Mental State Examination (MMSE; Folstein et al., 1975) is a pencil-and-paper screening test composed of 30 items assessing general cognitive efficiency (e.g., short-term and long-term memory, attention, visuo-motor coordination, spatial-temporal orientation). This tool provides a general cognitive efficiency score that according to the authors has to be corrected as a function of age and years of education. Internal consistency is expressed by a Cronbach's alpha ranging between .68 and .96. Test-retest reliability coefficients fall between .80 and .95. The level of sensitivity to correctly identify those individuals classified as cognitively impaired is .87, whereas the specificity coefficient which is used to identify those individuals who earlier were classified as cognitively intact is .82 (see Tombaugh & McIntyre, 1992).
The interview developed by Fastame and Penna (2012) was presented to collect information on lifestyle (e.g., time spent for outdoor leisure activities or gardening, type of hobbies, intake of medicine) and sociodemographic characteristics (e.g., age, years of education, marital status, people with which the respondents live).
Self-assessed personal satisfaction about family relationships index is a 10-point Likert-type scale containing two items assessing how much during the previous week the respondent was satisfied about his/her ties with family members. Zero indicated lack of satisfaction, whereas 10 denoted maximum satisfaction. The average score between the answers provided to the two items was calculated.
Self-assessed personal satisfaction about friendships index is a two-items measure asking participant to rate how much he or she was satisfied about his or her friendships and relations with the neighbors during the previous week along a 10-point Likert scale. The procedure used to calculate this index is identical to that adopted to rate the level of personal satisfaction about family relationships.
The CES-D (Radloff, 1977;Italian adaptation, Fava, 1983)  old is 43.62 (SD = 5.74), the mean score for the middle-aged (i.e., 43-57 years old) group is 41.23 (SD = 4.5), whereas the mean score for the over 57 years old is 43.8 (SD = 6.7). In the current sample the internal consistency reliability is expressed by a Cronbach's alpha of .73.

Procedure
Each participant was interviewed individually in a quiet room of his/her home. To minimize fatigue among older people, all participants responded verbally. All instructions and statements contained in the questionnaires used in the current study were read aloud by the third author who also recorded responses. Typically testing sessions lasted ~60 minutes.
Within each family group, the oldest participant (i.e., grandparent) first completed the MMSE. If the score indicated no cognitive decline (≥ 24), the socio-demographic interview was presented. Following this the order of the remaining measures was counterbalanced according to a Latin square procedure. This was repeated for the parents and grandchildren. If signs of cognitive decline were found in any participant, the entire family triplet was excluded by the study. This resulted in the exclusion of only one triplet and was due to mild cognitive decline in the grandparent.
The majority (90.4%) of participants reported sharing daily life with other family members (i.e., living with partner or parents or children vs. living alone). Specifically, 81% of the grandparents and 95% of the parents lived with their partner or children, and 95% of the young group lived with their partner or parents.
Intake of medicines (i.e., yes vs. no) was not counterbalanced across the three family groups, χ 2 (2, N = 63) = 22.23, p < .001. As expected, the use of drugs increased as a function of the age of the participants. Regular use of medicines was reported by 19% of young adults, 75% of their parents and 90.5% of their grandparents.
Attendance of socially-oriented activity (i.e., yes versus no) was counterbalanced across the participants, χ 2 (1, N = 63) = 2.70, p = .10. Overall, 62.3% of the respondents reported involvement in recreational/cultural or sport activities involving their social network. The breakdown across age groups was as follows -66.7% of grandchildren, 76.2% of their parents and 38.1% of their grandparents. Sporting activities were the main pastime of grandchildren (i.e., 57.1%) and parents (i.e., 47.6%) relative to participation in recreational/cultural activities (i.e., 9.5% and 28.6%, respectively). As expected, grandparents reported higher participation in recreational/cultural activities (i.e., 23.8%) relative to sporting ones (i.e., 14.3%). A high proportion (73%) of participants reported involvement in gardening/farming activities, χ 2 (1, N = 63) = 13.35, p < .001. This was expected considering agro-pastoral nature of this region of Sardinia. As participants were enrolled within the same families, the degree of intra-family correlations between the various mental health indexes (i.e., psychological well-being, personal satisfaction about family and non-family ties, depression) were calculated in terms of Intraclass Correlation Coefficients (ICC), using the methods suggested by Donner and Koval (1980). Asymptotic variance for the estimate of ICCs was also obtained.
Next, a multilevel modeling approach was used to investigate whether family membership (i.e., grandchildren, parents or grandparents enrolled within the same family) predicted self-reported psychological well-being (i.e., WEMWBS and CES-D, respectively), while controlling for the effect of marital status (i.e., single vs. engaged/ married), personal satisfaction about family relationships and personal satisfaction about non-family relationships. A random intercept was included, assuming simply that intercepts for the relationships between family members and psychological well-being (when controlling for marital status, and personal satisfaction about family relationships and personal satisfaction about non-family relationships) vary over families. The linear model also included random slopes, because it was hypothesized that psychological well-being scores could vary across grandchildren, parent and grandparent groups. Maximum likelihood was the method of estimation used to produce the parameter estimates.
The relationship between family membership and WEMWBS scores did not show any significant variance in intercepts across participants when the effect of the covariates was controlled for, var(u0j) = 2.07, χ 2 (1, N = 63) = 0.00, p > .05. Therefore, one can conclude then that the intercepts for the relationship between family membership and perceived psychological well-being (when controlling for the impact of personal satisfaction about family and non-family relationships and marital status) did not vary significantly across the different families. In addition, the slopes did not vary significantly across participants, var(u1j) = 2.68, χ 2 (1, N = 63) = 0.00, p > .05. Moreover, adding the covariance between slopes and intercepts did not make a significant difference to the model, cov(u0j, u1j) = 1. predicted WEMWBS index after controlling for the effect of personal satisfaction about family and non-family relationships and civil status, b = -1.21, t(63) = -2.24, p = .03. The same analyses were replicated using CES-D score as dependent variable. The relationship between family membership and depression did not show significant variance in intercepts across participants, var(u0j) = 9.15, χ 2 (1, N = 63) = 1.82, p > .05, suggesting that similar CES-D scores were found across the families. In addition, the slopes did not vary across participants, var(u1j) = 5.12, χ 2 (1, N = 63) = 0.00, p > .05, and the slopes and intercepts did not significantly

Discussion
The main aim of the current study was to explore whether family membership (i.e., grandchildren versus parents and grandparents) is a predictor of various measures of mental health (i.e., psychological well-being and depressive signs) in Sardinian families residing in the 'blue zone' of Arzana (Poulain et al., 2004), while controlling for the effect of marital status and perceived personal satisfaction about family and non-family relationships. Overall, several meaningful findings emerged from this preliminary investigation. First, considering that the respondents were raised in the same context (i.e., families living at Arzana), the calculation of ICC provided meaningful information, because it represents the proportion of the total variability in the outcome that is attributable to the families. With respect to the assessment of several mental health self-reported measures (i.e., depression, psychological well-being, personal satisfaction about family and non-family relationships) ICC values (i.e., < .30) reflected a marginal degree of similarity among participants enrolled within the same families (see Cohen,1988). Then, taking into account that family context had a little effect on the respondents, an HLM method was warranted to deal with the nested design of the dataset, since each participant belonged to one unique family. Therefore, the people tested for the study was the Level 1 variable, whereas the family to which they belonged represents the Level 2 variable. Overall, considering the main outcomes coming from the proposed models, one can conclude that, as suggested by the wider literature (e.g., Fuller-Iglesias et al., 2015;Poon & Cohen-Mansfield, 2011), the current findings document that apart from marital status, personal satisfaction about exchanges with family and non-family members, family membership played a crucial role for the assessment of WEMWBS score. Indeed, young adults reported greater life satisfaction than their grandparents. In contrast, CES-D score was significantly predicted only by the level of personal satisfaction about family and non-family relationships. Moreover, as hypothesized, similar level of satisfaction about family and non-family relationships were found across the three family generations (e.g., Kwan et al., 1997;Kitayama et al., 2000).
Various findings are also consistent with the lifestyle information provided by the respondents. Most participants attended socially oriented leisure activities (i.e., recreational/cultural and sport practices), that represent a good opportunity to establish or strength their friendships, reduce loneliness and improve physical activity. Such outcomes are known to be beneficial to mental health (Cornwell & Laumann, 2015). Furthermore, in agreement with previous studies conducted in the Sardinian 'blue zone' (e.g., Fastame, Penna, Rossetti, & Agus, 2014), an active lifestyle was characteristic of the sample and few participants were sedentary. In addition to spending time in socially oriented activities, the participants were often involved in gardening or farming. As suggested by previous authors, gardening has a positive impact on psychological well-being in adulthood (Heliker et al., 2001) and high participation is characteristic of rural areas of Ogliastra .
Moreover, in line with previous findings (e.g., de Jong Gierveld & Van Tilburg, 1999;Hank, 2007;Kitayama et al., 2000;Kwan et al., 1997) the current outcomes also show that family support promoted psychological well-being across adult life span (Dykstra, 2009;Margrett et al., 2011 To summarize, the present observations build on previous reports that superior mental health among inhabitants of the Sardinian 'blue zone' can be traced to the impact of various psychosocial factors (e.g., Fastame & Penna, 2012;. The present study identifies a novel effect, arising from close, satisfying interpersonal relationships and suggests this has a pervasive influence across adult life span (e.g., Carpiniello et al., 1989;Fastame, Penna, Rossetti, & Agus, 2014). However, this is a very preliminary investigation, therefore it presents some limitations. First, the sample size is small and the participants were recruited only in the Sardinian 'blue zone'. Although the tendency to look after and support elderly people is typical of Mediterranean countries (e.g., de Jong Gierveld & Van Tilburg, 1999;Hank, 2007) further investigation with larger samples of differently aged adults from other regions is necessary. This would help determine whether highly context-specific combinations of psychosocial factors, including the characteristics connected with family and non-family networks, promote successful aging within the Sardinian 'blue zone'. However, if the current findings can be generalized, it would suggest that this population may represent a valuable resource for the investigation of successful ageing. Second, to our knowledge it is the first involving use of the WEMWBS scale among Italian the oldest-old (i.e., over 80 years old) participants. An obvious limitation is that the stringent inclusion criteria constrained the number of families that could be recruited from such a small village (i.e., population < 2500 inhabitants). While the strict inclusion criteria enhance the reliability of the findings cautious interpretation of the findings is essential. Third, the impact of social desirability on the self-assessment of psychological well-being and depressive signs was not investigated. Future research has to address this issue, because a recent trend of research (Fastame & Penna, 2012;Fastame, Hitchcott, & Penna, 2017;Hitchcott, Fastame, Penna, & Agus, 2016) suggests that socially desirable responding style should be taken into consideration when self-rated mental health measures of mental health are administered, because social desirability scores tend to increase with age and in women.
In conclusion, replication of the current findings, would promote understanding of mental health and successful ageing and, possibly, shape health service policies aimed at promoting quality of life in Italian adults.

Funding
This work was supported by the Sardinia Regional Government under grant "CRP-78543" entitled "Invecchiamento attivo in Sardegna: quali fattori influenzano il benessere psicologico negli anziani?" [Active ageing in Sardinia: What factors influence psychological well-being in the elderly?]" and awarded to the authors except the third one.