In general, the findings of this study were consistent with those of the the previous
literature, showing that poor subjective and objective health, poor results in
various measures of physical functioning, low psychological well-being and
low perceptions of social support predicted mortality in older people. However,
some new findings emerged: 1) the association between self-rated health and
mortality was explained by diagnosed diseases, but not with performancebased
measures of functioning; 2) no single predictor of mortality from
performance–based measures of functioning emerged; 3) low life satisfaction
related to the present, in terms of zest and mood, was a more important
predictor of mortality than congruence representing the satisfaction with past
life; 4) the perceived non-assistance related social support was more important
predictor of mortality than perceived assistance-related social support in the
older women, and 5) social support was not related to mortality risk in older
men.
Self-rated health and mortality
Previous studies of self-rated health and mortality have shown that poor selfrated
health predicts subsequent death even when several sociodemocraphic
and health indicators are controlled for (Mossey & Shapiro 1982, Idler & Benyamini
1997, Benyamini & Idler 1999, Shadbolt et al. 2002). Nevertheless, in
most previous studies objective measures of health and functioning have not
been included in the analyses. When information about potential confounders is
based on self-reports, these confounders are likely to be influenced by individual
styles of responding and thus, be biased in the same direction as self-rated
health itself. Very few studies have included data from performance-based
measures, clinical examinations or laboratory analyses (Kaplan et al. 1996,
Simons et al. 1996, Jylhä et al. 2006).
46
In study I, conducted with a sample only containing men, self-rated health
was a significant predictor of mortality when performance-based indicators of
functioning were controlled for, but not when the number of diagnosed
diseases was included in the model. Jylhä et al. (1998) have argued earlier that
self-rated health is a personal, subjective summary of all the information people
have about anything they understand as elements of “health”. Jylhä et al.
(2006), on the basis of the findings that self-rated health showed a graded
association with frequently used biomarkers, concluded that self-rated health
also has a biological basis. When biomarkers were included in the model, selfrated
health still was a significant predictor of mortality.
There are also some evidence of gender differences in self-rated health and
mortality risk (Deeg & Kriegsman 2003), but a male-only sample did not allow
this issue to be studied in study I.
Physical functioning and mortality
Among the performance-based measures of functioning, poor mobility was one
of the strongest predictors of mortality (I,II). In earlier studies impaired mobility
has been shown to be a major risk factor for disability, falling, loss of independence,
hospitalization and mortality (Guralnik et al. 2000, Hirvensalo et al.
2000, Penninx et al. 2000, Guralnik et al. 2001, Stel et al. 2003).
Slow walking speed has been found to be related to several chronic
conditions and weakened muscle strength. In this study older people whose
muscle strength was weak, were at higher risk for death. Several other studies
have found also a similar association (Rantanen et al. 2000, Metter et al. 2002,
Snih et al. 2002, Rantanen 2003). The suggested pathways from disease to
muscle impairment include nutritional depletion, systemic inflammation, and
physical inactivity. Poor muscle strength could be a marker of disease severity,
which in turn is associated with mortality (Rantanen 2003). Decreased muscle
strength is also a risk factor for falls, which in turn may lead to higher mortality
risk. A recent study confirmed that strength is a more important predictor of
mortality in older people than muscle mass (Newman et al. 2006b).
Speed-related functioning, like slow tapping time has been found to be
associated with increasing mortality in older people independently of the
effects of muscle power and strength (Metter at al. 2005). Era (1987) found
higher means for tapping rate in 71-year-old male survivors than non-survivors
at a 5-year follow-up. These studies support the present finding that slow
tapping rate is related to higher risk for mortality.
In study II, low vital capacity was related to higher mortality risk. Earlier
studies have shown that reduced pulmonary function is an important predictor
of cardiovascular morbidity and mortality (Sorlie et al. 1989, Sharps et al. 1997,
Schunemann et al. 2000). Respiratory functions are good indicators of health
status and especially the coronary disease related mortality. The association
between pulmonary function and coronary heart disease can partly be
explained by cigarette smoking, which leads to both lung impairment and
incidence of coronary heart disease (Marcus et al. 1989). In present study this
47
was not the case, as in the initial analyses it was shown that controlling for
smoking did not reduce the hazard ratios. Also, in this sample of Finnish 75-
year-old people, the percentage of smokers was very low, especially in women.
The main finding of study II was that the functions measured were related
to higher mortality risk to a similar extent, and thus no single dominant
predictive factor emerged. It was possible to draw this conclusion, because
several performance-based measures were studied simultaneously.
Furthermore, the application of the principal component estimation procedure
in the Cox regression model yelded a truer picture of the impact of an
individual’s functional status and mortality. The levels of the various functions
studied showed significant intercorrelation, indicating that a decline in one
function also contains information about the other physical functions of an
individual. A study by Heikkinen et al. (1993) found that in older male age
group (71–75-years) intercorrelations in measures of functiong were higher than
in younger age groups.
It could be argued that the older people with higher risk for death
represented the frailty phenotype. Frailty is a robust concept with the risk of
adverse outcomes being largely established by age 70 (Rockwood et al. 2006).
Ferrucci et al. (2003) have put forward hypothesis on the development of frailty,
according to which disease, disuse and ageing “per se” trigger a mechanism
that exhausts the redundancy of muscular and nervous backup systems, so that
the damage goes beyond the threshold of possible compensation and leads to a
measurable decline in physical performance and in the end, to death.
At the other end of continuum there is a group of survivors, whose good
level of physical functions may represent general vitality or robustness, thus
protecting those individuals from premature death. The explanation for this
could be found in their earlier lifestyle and partly in genetic influences. Recent
studies have suggested that genetic influences contribute to about one third of
the variation in postural balance, muscle strength and mobility in older people
(Pajala et al. 2004, Tiainen et al. 2004, 2005, Pajala et al. 2005, 2006).
It also seems that the protective effects of good physical functions are
partly additive. When these functions were studied univariately, the risk for
death for those in the lowest tertile was high, and when gathered together in an
index the information of different functions showed an even higher risk for
mortality for these individuals.
Life satisfaction and mortality
Life satisfaction has mainly been used as an endpoint in earlier ageing research,
providing a lot of useful information about its determinants and predictors.
Studies of the impact of life satisfaction on various health outcomes in an older
population are few, although this issue is particularly important in old age
when various deficits in health, functional capacity and social network
accumulate. Although many studies have found other indicators of
psychological well-being than life satisfaction to be highly related to lower risk
for mortality in old age (Maier & Smith 1999, Ostir et al. 2000, Penninx et al.
48
2000b, Danner et al. 2001, Levy et al. 2002, Giltay et al. 2004, Pitkälä et al. 2004),
no studies using validated measures of life satisfaction and mortality in older
people were located.
Life satisfaction index Z (LSIZ) (Wood et al. 1969), a well-validated
measure and designed especially for older people, was applied in study III. In
order to study the factor structure of LSIZ and the mortality risk related to its
different subscales, exploratory factor analysis was conducted. The analysis
yielded three factors. The first factor was identified as Zest, and it represented
satisfaction with present life and zest. The second factor was identified as
Mood, and represented present life satisfaction and mood. The third factor,
Congruence, represented past life satisfaction and congruence. This factor
structure has earlier been found in Life satisfaction index A, which is the
original, larger version of LSIZ (Adams 1969, Hoyt & Creech 1983, Liang 1984,
Shmotkin 1991). From these three factors Zest and Mood, representing
satisfaction with present life, were associated to mortality risk. The Congruence
factor, representing satisfaction in past life and achieving personal goals, was
not related to higher mortality.
It seems that old-old people perceive their satisfaction with the life they
are now living as more important than issues that are usually considered
important in younger age groups. There are similar findings in earlier study of
the life-lines of older people by Takkinen and Suutama (2004). Socioemotional
selectivity theorists also argue that when people perceive their time in life to be
limited, futuristic goals become less relevant and present-oriented, emotional
goals become more important (Carstensen et al. 1999, Cheng 2004).
Studies showing that optimistic attitude and feelings of hope are related to
better survival in old age (e.g. Giltay et al. 2004), support our finding. Positive
emotions may promote a more active social and physical lifestyle and also
motivation toward self-care (Scheier & Carver 1992). Older people with better
subjective well-being may also have better coping capabilities; optimists usually
avoid strategies like denial and giving up (Segerström et al. 1998).
Social support and mortality
Earlier studies on social support and mortality have mostly focused on
quantitative parts of social network (e.g. Avlund et al 1998). In study IV,
perceived social support was measured with well-validated instrument, the
Social Provision Scale (Weiss 1973, Russell et al. 1984, Cutrona & Russell 1987,
Törmäkangas et al. 2003), which measures the kind of support a person gets
from his or her social network.
Although there is much evidence of beneficial effects of social support on
the health of older people, the different subscales and their association with
mortality have not been studied earlier. Thus, the items of Weiss’s Social
Provision Scale were divided into two categories: assistance-related and nonassistance-
related (Cutrona & Russell 1987). Perceptions of poor non-assistancerelated
perceived social support, consisting of reassurance of worth, emotional
closeness, sense of belonging and opportunity for nurture, predicted mortality of
49
older women. Assistance-related social support, consisting of guidance (advice
or information) and reliable alliance (assurance that there are people who can
be counted on in certain circumstances) was not related to survival in the
women. This finding is in line with earlier studies showing that emotional
social support is more important to the health of older people. In fact, there
have been some studies showing that receipt of instrumental social support is
unfavourable for health outcomes, probably because underlying poor health
status that requires instrumental support; favorable effects have been mainly
demonstrated for emotional support (Seeman et al. 1996, Penninx et al. 1997).
Identity theorists argue that being imbedded in a social network is
protective because it gives the individual meaningful roles; these in turn
generate self-esteem and a sense of purpose in life (Thoits 1983, Unchino 2004).
Good social relationships also help older people to cope with different losses
that ageing may bring alone (Akiyama & Akiyama 1991). The absence of close
attachments and recognition of worth causes emotional loneliness and
depressiveness (Heikkinen & Kauppinen 2004, Taylor & Lynch 2004). The
feeling of being needed and valued is important; it gives one the strength to
take care of oneself. Those who are alone and forgotten, despite formal support,
are at a higher risk for death. Older women themselves also connect good social
relationships with their experienced health (Lyyra & Heikkinen 2006).
Differences between age groups
In the final models of study I the significance of certain predictors of mortality
varied between the middle-aged and older men: in the middle-aged group high
serum total cholesterol predicted mortality. In the older age group the variables
indicating poor physical strength and cognitive performance were associated
with higher mortality, indicating that the predictors of mortality in older people
are different from those predicting mortality in middle-aged people.
Low body mass index was associated with a higher mortality risk in the
older age group. Consistently with the studies by Grabowski and Ellis (2001)
and Landi et al. (1999) the result of this study suggested that a higher body
mass index might be protective in older people. In study I body mass index in
middle age was not related to mortality over an 18-year-follow-up. It is possible
that longer follow-up would have produced a different result, as in the study by
Yan et al. (2006), where the mean follow-up was 32 years; they found that those
who were obese in middle age had a higher risk for hospitalization and
mortality in older age than those who were normal weight.
A similar phenomenon was found regarding to serum total cholesterol. In
study I, higher serum total cholesterol was associated with increased mortality
in the middle-aged but not in older men. The results of several studies have
shown either an inverse or no relation between total cholesterol and mortality
in older people (Pijls et al. 1993, Krumholz et al. 1994, Chyou & Eaker 2000,
Brescianini et al 2003). One explanation could be that the men at highest risk
may already have died prior to the study. Some findings also suggest that
between older and younger populations the cholesterol metabolism may differ
50
(Chyou & Eaker 2000). Menotti et al. (1998) have found that out of 11 powerful
cardiovascular risk factors predicting mortality only five remained significant
when measurements were done 10 years later for 50-year-old men, suggesting
that their predictive power declines with advancing age. There is ongoing
debate on the use of lipid-lowering drugs in older populations and obviously
more randomised controlled trials are needed.
Differences between sexes
The pattern of the associations between physical functions and mortality was
similar in both sexes (II). The only difference was found in vibration threshold,
which was related to mortality in the 75-year-old women, but not in men.
Higher vibration threshold has been found to associate with several chronic
conditions, such as diabetes. Vibration threshold also correlated strongly with
number of chronic diseases in the women in this sample.
A significant association between mortality and social support was found
only in the 80-year-old women (IV). This is contradictory to the findings of
earlier studies (House et al. 1982, Kaplan et al. 1988). Studies performed
exclusively with older people have found that social support predict survival in
both sexes (Jylhä & Aro 1989, Seeman et al. 1993). Men tend to maintain close,
intimate relationships with only a few people, primarily with their spouses.
Being married has been found to be more protective for men (Berkman & Syme
1979). In this study the majority of the men were married, whereas most of the
women were widowed and lived alone. For the women living alone the quality
of social relationships seems to be more important than it is for men who are
married. Earlier studies have also shown that in women the social network and
contacts increase after the death of a spouse (Barrett & Lynch 1999, Lennartsson
1999, Pinquart 2003). In men, the situation is the reverse: becoming widowed
leads to a decrease in social contacts (van Grootheest et al. 1999).
