Highlights
• Happiness and generalised trust both touted as independent predictors of mortality.
• Trust but not happiness predicts all-cause mortality.
• Trust predicts mortality caused by CVD but not by neoplasia.
• Psychosocial mechanisms might drive the association between trust and health.
Abstract: There has been an abundance of research discussing the health implications of generalised trust and happiness over the past two decades. Both attitudes have been touted as independent predictors of morbidity and mortality, with strikingly similar trajectories and biological pathways being hypothesised. To date, however, neither trust nor happiness have been considered simultaneously as predictors of mortality. This study, therefore, aims to investigate the effects of generalised trust and happiness on all-cause and cause-specific mortality. The distinction between different causes of death (i.e. cardiovascular vs. cancer-related mortality) allowed us to assess if psychosocial mechanisms could account for associations between generalised trust, happiness and mortality. The study sample was derived from US General Social Survey data from 1978 to 2010 (response rates ranged from 70 to 82 per cent), and combined with death records from the National Death Index. The analytical sample comprised 23,933 individuals with 5382 validated deaths from all-cause mortality by 2014. Analyses were performed with Cox regression models and competing-risk models. In final models, generalised trust, but not happiness, showed robust and independent associations with all-cause mortality. Regarding cause-specific mortality, trust only showed a significant relationship with cardiovascular mortality. The distinct patterns of association between generalised trust and all-cause/cause-specific mortality suggest that their relationship could be being driven by cardiovascular mortality. In turn, this supports the feasibility of psychosocial pathways as possible biological mechanisms from distrust to mortality.
Keywords: TrustHappinessAll-cause mortalityCause-specific mortalityPsychosocial pathwayCox regressionCompeting-risk regressionUnited States
Discussion
This US population-based study is the first to investigate whether individual-level generalised
trust and happiness independently predicted all-cause mortality. By using cause-specific
mortality outcomes, we further sought to corroborate the hypothesis that psychosocial
mechanisms could provide a feasible pathway from low trust and unhappiness to mortality.
Our findings confirmed that individual-level trust maintained independent and robust
associations with all-cause and cardiovascular-specific mortality, even after socio-economic
and other demographic factors were considered. Results presented here thus mirror previous
empirical findings of associations between generalised trust and longevity (Islam et al., 2006;
Kawachi et al., 1997; Murayama et al., 2012).
Conversely, associations between happiness and all-cause mortality were fully attenuated
once adjusting for confounders. Several mechanisms have been proposed as to why
generalised trust may lead to better health and longevity. Some argue that trust mobilises
social support and enables greater collective action, providing greater access to those
resources needed to cope better with any potential health hazard (Elgar, 2010; Moore &
Kawachi, 2017). Others hint at the genetic aetiology of trust (Oskarsson et al., 2012; Wootton
et al., 2016), though there are currently no studies that investigate if the genetic underpinnings
of distrust/trust are also linked with known disease risk and/or protective gene variants.
That generalised trust is robustly associated with cardiovascular mortality in this study lends
further weight to the idea that psychosocial pathways are a plausible biological mechanism
from trust to health (Abbott & Freeth, 2008). To clarify further, if individual-level trust
reflects social trustworthiness, then lower levels of trust could be indicative of higher social
stressors (Giordano & Lindström, 2010; Wilkinson, 1996). From this perspective, generalised
trust acts as buffer reducing the anxiety stemming from the behaviour of others (Abbott &
Freeth, 2008). If high trust facilitates collective action (Coleman, 1988), then it is reasonable
to assume that low trust hinders this process, creating greater concern during every-day
transactions compared to those conducted within a ‘high-trust’ milieu. It has been argued that
exposure to high levels of social stressors could have a deleterious impact on the
cardiovascular system. The biological pathways through which this acts is the hypothalamic
pituitary-adrenal (HPA) axis, the overstimulation of which leads to increased levels of blood
cortisol (Rosmond & Björntorp, 2000). Prolonged and/or repeatedly high blood cortisol levels
released under conditions of perceived stress have been shown to increase one’s risk of
atherosclerosis (Dekker et al., 2008) and coronary artery calcification over the life-course
(Hamer et al., 2012).
In this study, individuals who distrusted others had, in comparison to the trusting group, a 13
per cent elevated risk of death caused by cardiovascular disease (Table 3, Model 7 & 8).
However, from the analyses performed, it is not possible to distinguish if individual trust is an
interpretation of environmental trustworthiness (hinting at the social capital debate) or
whether it captures pathological distrust, an element of the personality trait known as cynical
hostility (Kawachi, 2018). Cynically hostile individuals are also reported to have an increased
cardiovascular mortality risk, with possible pathways from distrust to cardiovascular mortality
including socio-economic status and the same HPA-axis mechanisms previously described
(Everson et al., 1997).
Strengths & Limitations
This is the first study to independently investigate the effects of both generalised trust and
happiness on mortality outcomes, using rich US survey data that span over more than three
decades. The GSS data were prospectively linked to mortality registries from the NDI, which
provided objective and validated specific cause-of-death categories. While these pooled GSS
data are nationally representative, this study design relied on single cross-sectional
observations, which do not capture change over time. Though a study based on UK panel data
showed how individuals’ generalised trust could change (Giordano et al., 2012), individuals’
generalised trust tends to re-adapt to a certain ‘set point’ in the longer term (Dawson, 2017).
Considerable stability is also attributed to levels of happiness (Lucas & Donnellan, 2007).
While our study corroborates hypotheses linking generalised trust to longevity, our analysis
has consciously ignored important parts of the wider debate on social capital and health. We
refrained from analysing additional social capital measures for three important reasons. The
first is a simple methodological one: not enough rounds of the GSS contained the desired
measures to obtain statistically powerful samples. Secondly, while generalised trust is ‘sticky’
in adulthood (Uslaner, 2002), other important social capital proxies (e.g. membership in
(voluntary) associations and community ties are not. Our data lacked the possibility to track
individuals’ membership, networks and community social capital longitudinally, making
inferences from any estimates untrustworthy. Thirdly, using Canadian survey data, Carpiano
and Fitterer (2014) have showed that generalised trust could be conceptually different from
other social capital measures.
While survey research generally favours multiple-item scales over single-item measures, our
measures of trust and happiness belonged to the latter group. Regarding happiness, the GSS
simply lacks additional measures. As for trust, previous research highlighted that the single
item trust measure outperforms the GSS three-item trust scale in terms of reliability and
validity (Uslaner, 2015). Moreover, the standard single-item trust measure has, for a long
time, featured in a range of international survey studies. Opting for the single-item trust
measure thus increases the possibility of replication in future studies in other contexts.
We investigated cause-specific mortality in an attempt to substantiate that psychosocial
pathways were one plausible biological mechanism from generalised trust to health.
Unfortunately, there is no possibility to track health behaviour in the GSS-NDI data after
1994, as questions regarding smoking and drinking were no longer employed. We thus lacked
the opportunity to establish associations between trust and CVD mortality adjusting for risky
health behaviour. We deliberately focused on deaths caused by either CVD or by neoplasia
for two reasons. Firstly, because psychosocial pathways are purported to play a greater role in
CVD-related deaths. Secondly, they are the two most frequent causes of death in these data.
Future studies could investigate other associations between trust and cause-specific deaths,
e.g. the infamously theorised association between (a lack of) generalised trust and suicide
(Durkheim, 2005). Unfortunately, the GSS-NDI drawn for purposes of this study simply lack
the statistical power to consider further categories of cause-specific mortality. Finally, all
analyses were conducted at the individual level, which makes it impossible to ascertain
whether presented relationships with mortality are due to trust being an individual or a
contextual resource (Giordano et al., 2019).
Posts
No comments:
Post a Comment