Sei sulla pagina 1di 12

Social Science & Medicine 66 (2008) 2401e2412

www.elsevier.com/locate/socscimed

Comparing sense of coherence, depressive symptoms and anxiety,


and their relationships with health in a population-based study*
Hanna Konttinen a,*, Ari Haukkala a, Antti Uutela b
a
Department of Social Psychology, University of Helsinki, Unioninkatu 37, P.O. Box 54, 00014 Helsinki, Finland
b
National Public Health Institute, Finland
Available online 11 March 2008

Abstract

The strong negative correlations observed between the sense of coherence (SOC) scale and measures of depression and anxiety
raise the question of whether the SOC scale inversely measures the other constructs. The main aim of the present study was to
examine the discriminant validity of the three measures by comparing their associations with health indicators and behaviours.
The participants were 25 to 74-year-old Finnish men (n ¼ 2351) and women (n ¼ 2291) from the National Cardiovascular Risk
Factor Survey conducted in 1997. The SOC scale had high inverse correlations with both depression (r ¼ 0.62 among both
men and women) and anxiety measures (r ¼ 0.57 among the men and r ¼ 0.54 among the women). Although confirmatory
factor analyses suggested that it was possible to differentiate between SOC, cognitive depressive symptoms and anxiety, the
estimated correlations were even higher than those mentioned above. Education was related only to SOC, but the associations
of SOC, cognitive depressive symptoms and anxiety with self-reported and clinically measured health indicators (body mass index,
blood pressure, cholesterol) and health behaviours were almost identical. The variation in the lowest SOC tertile was more strongly
associated with health variables than in the highest tertile. To conclude, the size of the overlap between the SOC and depression
scales was the same as between depression and anxiety measures. This indicates that future studies should examine the discriminant
validity of different psychosocial scales more closely, and should compare them in health research in order to bring parallel
concepts into the same scientific discussion.
Ó 2008 Elsevier Ltd. All rights reserved.

Keywords: Finland; Sense of coherence; Depression; Anxiety; Health; Psychological measurement; Discriminant validity; Mental health

Introduction emotions (e.g., hostility, depression, anxiety) hypothe-


sized to affect disease onset and development (Suls &
Personality and emotional factors have long been Bunde, 2005), as well as, positive psychosocial vari-
suggested to have an influence on physical health. ables (e.g., optimism, social support, sense of coher-
Research efforts have been directed to the negative ence) that are assumed to maintain and enhance
health. The tendency among researchers has been to
*
Support to conduct this research was provided by the Signe and evaluate the independent effects of different psychoso-
Ane Gyllenberg Foundation. cial factors on physical health, while ignoring possible
* Corresponding author. Tel.: þ358 919124878.
E-mail addresses: hanna.konttinen@helsinki.fi (H. Konttinen),
overlap among variables (Kaplan, 1995). Sense of
ari.haukkala@helsinki.fi (A. Haukkala), antti.uutela@ktl.fi coherence (SOC) attracted a lot of attention after the
(A. Uutela). development of a quantitative scale to measure it.

0277-9536/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.socscimed.2008.01.053
2402 H. Konttinen et al. / Social Science & Medicine 66 (2008) 2401e2412

Nevertheless, researchers such as Geyer (1997) have SOC reduces the health-damaging effects of stress by
criticized the ambiguity of the scale and its lack of dis- lowering the probability of adverse physiological
criminant validity in relation to the measures of depres- reactions and negative emotions associated with stress
sion and anxiety. perceptions. Antonovsky (1987) emphasized that the
consequences of stressors may be negative, neutral or
The sense of coherence construct positive, depending on their nature and the adequacy
of coping. It is also possible that persistent or serious
Antonovsky (1979, 1987, 1993) proposed a saluto- health problems may influence the development of
genic orientation focusing on factors that protected SOC.
individuals’ health. He defined the central concept, Numerous cross-sectional studies have found SOC
SOC, as an orientation towards life that characterises to be positively associated with physical and psycho-
the extent to which an individual appraises internal logical health (Flannery & Flannery, 1990; Larsson
and external environments as (1) comprehensible, (2) & Kallenberg, 1996; Lundberg, 1997; Pallant & Lae,
manageable and (3) meaningful. Antonovsky (1987) 2002; Suominen, Blomberg, Helenius, & Koskenvuo,
developed these three interrelated dimensions and the 1999), and health behaviours (e.g., alcohol problems
SOC questionnaire on the basis of interviews with and physical activity) (Kuuppelomäki & Utriainen,
persons who had experienced severe trauma. He 2003; Midanik, Soghikian, Ransom, & Polen, 1992).
maintained that SOC has much in common with the There have been a few longitudinal studies investigat-
psychological constructs of hardiness (Kobasa, 1982), ing whether SOC predicts health outcomes (e.g., all-
locus of control (Rotter, 1966) and self-efficacy cause mortality and subjective state of health). Most
(Bandura, 1977). of these have supported its predictability (Suominen,
The SOC scale exists in two forms: the original Helenius, Blomberg, Uutela, & Koskenvuo, 2001;
29-item questionnaire and the shortened version Surtees, Wainwright, Luben, Khaw, & Day, 2003),
including 13 items from the full form. Antonovsky but inconsistent results have also been reported.
(1987, p. 87, 1993) designed each item to represent Kivimäki, Feldt, Vahtera, and Nurmi (2000) found in
one of the three dimensions, but he emphasized that their study of municipal employees that SOC predicted
the scale measured the SOC construct unidimension- sickness absence only among the women. Other
ally. The structure of the SOC measure has since research among people with chronic illness reported
been investigated using both exploratory and confirma- reciprocal causation between SOC and domains of
tory factor analysis, but with inconsistent results. Some health (Veenstra, Moum, & Roysamb, 2005). Overall,
studies employing exploratory factor analyses have the SOC relationship with psychological well-being
supported the one-factor structure proposed by seems to be stronger and more direct than that with
Antonovsky (e.g., Frenz, Carey, & Jorgensen, 1993), physical health (Eriksson & Lindström, 2006).
but other results have also been obtained (e.g., Sandell,
Blomberg, & Lazar, 1998). Furthermore, a confirma- Depressive symptoms and anxiety
tory factor analysis study on the 13-item SOC
questionnaire conducted among Finnish working-aged The conceptual and empirical relationship between
people found support for the three-factor model (Feldt depression and anxiety has been studied extensively,
et al., 2006). Klepp, Mastekaasa, Sorensen, Sandanger, and the evidence suggests that it is difficult to differen-
and Kleiner (2007), on the other hand, preferred the tiate these two constructs empirically. The emotion of
one-factor solution for a brief nine-item SOC scale fear has a central role in anxiety, which involves
because the factors in the two- and three-factor models feelings of worry, apprehension and dread (Watson &
were very strongly correlated, and the factor scores had Kendall, 1989). The essential characteristic of depres-
similar correlations with measures of psychological sion is sadness, with associated feelings of sorrow,
well-being, depression and anxiety. hopelessness and gloom. Self-reported depression and
The SOC theory hypothesizes that an individual anxiety have been highly correlated (typically between
with a strong SOC maintains and enhances health 0.45 and 0.75) in both psychiatric and non-psychiatric
through effective and flexible coping with stressors, samples (Clark & Watson, 1991). Considerable co-
such as by adopting health-enhancing and avoiding morbidity has also been observed (Mineka, Watson,
unhealthy behaviours (Antonovsky, 1987). An individ- & Clark, 1998).
ual with a strong SOC is also more likely to perceive The strong relationship between depression and
internal and social environments as non-stressful. anxiety may reflect problems with existing scales
H. Konttinen et al. / Social Science & Medicine 66 (2008) 2401e2412 2403

and constructs. There is an overlap in many of the The relationship of SOC with depressive
symptoms that define both syndromes, and many symptoms and anxiety
self-report scales also contain items that actually
measure the other construct (Gotlib & Cane, High negative associations have been consistently
1989). Thus, Clark and Watson (1991) proposed a tri- observed between the SOC scale and measures of
partite model of depression and anxiety: general depressive symptoms and anxiety. In studies published
distress or negative affect is common to both, while between 1992 and 2003, the correlation coefficients
manifestations of low positive affect are specific to between the scale and different measures of anxiety
depression, and symptoms of somatic arousal signify and depression varied from 0.29 to 0.82 and
anxiety. from 0.34 to 0.90, respectively (Eriksson & Lind-
The two best-known theories of depression are ström, 2005). Gruszczyńska (2006) calculated the
perhaps Beck’s (1967, 1987) cognitive theory and mean weighted SOC correlations on the basis of 17
Abramson, Metalsky, and Alloy’s (1989) hopelessness studies: 0.70 for anxiety and 0.65 for depression.
theory. They both posit that cognitive vulnerability and The size of these correlations raises doubts about the
negative life stress together precipitate depression. construct validity of the SOC questionnaire. Thus, it
According to Beck (1987), dysfunctional attitudes has been proposed that rather than being a measure
involving feelings of loss, inadequacy, failure and of resilience, the scale measures inversely negative
worthlessness constitute cognitive vulnerability to affectivity/neuroticism (Frenz et al., 1993; Strümpfer,
depression, whereas in the view of Abramson et al. Gouws, & Viviers, 1998; Watson & Clark, 1984).
(1989), cognitive depression diathesis encompasses There are a few studies that have investigated the
the general tendency to attribute negative events to sta- relationships between SOC, anxiety, and depression.
ble and global causes, to infer negative consequences, When confirmatory factor analysis was used, the 29-
and/or to infer negative characteristics about the self. item SOC scale loaded on the same latent variable as
In addition, Beck (1987) postulates that all depressed the measures of trait anxiety, neuroticism, optimism,
people show a cognitive triad: automatic thoughts depression and self-efficacy (Gruszczyńska, 2006),
reflecting negative views of the self, the world and and also on both the health-proneness and negative-
the future. It is suggested that the degree of this affect factors (Kravetz, Drory, & Florian, 1993). Feldt,
negative thinking is directly related to the severity of Metsäpelto, Kinnunen, and Pulkkinen (2007) found
other depressive symptoms. that a high SOC (measured on the 13-item question-
The SOC construct could be integrated as a protec- naire) was strongly associated (r ¼ 0.86) with
tive/risk factor into vulnerability-stress models of reversed neuroticism in their confirmatory model.
depression. Individuals with a strong SOC should be Korotkov (1993) concluded from his four-week
better able to maintain their emotional well-being in prospective study that the 13-item SOC instrument
stressful situations, whereas those with a weak SOC lacked face, construct and predictive validity, and
share some of the cognitive vulnerability factors men- that most of the items measured emotionality (neurot-
tioned in Abramson et al.’s (1989) and Beck’s (1967) icism) rather than SOC. Meanwhile, Breslin, Hepburn,
theories. All this raises a question concerning vulnera- Ibrahim, and Cole (2006) recently analysed the longi-
bility and protective factors in general: do they form tudinal relationship between psychological stress and
different dimensions or are they merely opposite ends the 13-item SOC scale, and found that the stable
of a continuum? Antonovsky (1987) considered risk components of distress and SOC were strongly inter-
and protective factors qualitatively different. The correlated (r ¼ 0.86). On the other hand, Strümpfer
SOC construct also shares similarities with cognitive et al. (1998) conceptualised the low end of negative af-
and emotional features of depression, which raises fectivity as emotional stability, and interpreted the
the question of whether a weak SOC represents depres- strong negative associations between both versions of
sion. Overall, SOC describes individuals’ cognitive the SOC instrument and the negative-affectivity scales
and emotional functioning on a more general level as supporting the validity of the instrument. However,
than theories of depression. In addition, the SOC the SOC measures were more strongly associated
theory is more oriented towards societal determinants with negative than with positive emotionality.
(e.g., education) of health than depression theories. In sum, the few previous studies focusing on the
There is evidence that people in higher socio-economic relationship between the SOC scale and measures of
positions have a stronger SOC (Lundberg & Nyström negative emotions have yielded inconclusive results.
Peck, 1994). SOC, depressive symptoms and anxiety have all been
2404 H. Konttinen et al. / Social Science & Medicine 66 (2008) 2401e2412

related to various health indicators and behaviours, but Helsinki metropolitan area, respectively. Two-thirds
to our knowledge their associations with health-related of these participants were randomly assigned to the
variables have not been compared in any previous present study of psychosocial factors (n ¼ 7103).
study. The purpose of the present study therefore was The participants received by mail a self-adminis-
to investigate the discriminant validity of the SOC, tered health questionnaire including questions on soci-
depression and anxiety scales in a population-based odemographic factors (gender, education and income),
sample. In more specific terms: (1) first, we examined health behaviours (smoking and alcohol consumption),
the strength of the correlations between the measures and health (self-rated health, visits to a physician,
of SOC, depressive symptoms and anxiety, and con- somatic and psychological symptoms). They were
firmed these with a confirmatory model that takes mea- asked to fill in the questionnaire at home, and to return
surement error into account. (2) Our main aim was to it to the health centre when they came for a medical
assess the discriminant validity between SOC, depres- examination in which their weight, height and blood
sion and anxiety scales by comparing their associations pressure were measured, and blood samples were
with subjective and objective (body mass index, blood taken. A self-administered psychosocial questionnaire
pressure, total cholesterol) health indicators, health was then given to a sub-sample of these participants,
behaviours and sociodemographic factors. If the SOC to be answered afterwards at home. This comprised
scale and the measures of depressive symptoms and a 13-item SOC questionnaire followed by a 21-item
anxiety are distinct measures, they should also be depression inventory and a six-item anxiety inventory.
differentially related to other variables. (3) Finally, All three scales were separated by a number of other
we wished to find out whether variation in the SOC items. Only the respondents who participated in all
was more strongly related to health variables at the three phases were included in the present study. The
high or low end of the distribution. The questionnaire final response rate was 67% (men, n ¼ 2351; women,
produces variation among individuals with a weak n ¼ 2291).
and a strong SOC. Most depression inventories, on
the other hand, do not account for variation among Measures
the non-depressed as they assess only the intensity of
depressive symptoms. Antonovsky specifically hypoth- Sense of coherence (SOC)
esized that a strong SOC, and not just the absence of We used a Finnish-language version of Antonov-
weak SOC, protected health. If variation at the high sky’s (1987) 13-item orientation to life questionnaire
end were associated with health measures, then this consisting of four meaningfulness, five comprehensi-
would give the SOC scale an advantage over depres- bility, and four manageability items to measure SOC
sion measures. (for the items and their order in the questionnaire,
see Fig. 1). The respondents were asked to indicate
Methods the number that expressed their feeling best on
a seven-point scale with two anchoring phrases (e.g.,
Participants 1 ¼ very often, 7 ¼ very seldom or never). As recom-
mended initially by Antonovsky (1993) and recently
The participants comprised a random sub-sample of by Feldt et al. (2006), one total SOC score was calcu-
the national cardiovascular risk factor survey (FIN- lated from the ratings of the 13 questions. The SOC
RISK) conducted in Finland in 1997 (Vartiainen scores varied from 16 to 91, the higher scores reflecting
et al., 2000). FINRISK 97 covered a random sample stronger SOC. The scale was internally consistent on
of 11,500 people aged 25e74 drawn from the Finnish the basis of the Cronbach’s alpha criterion (0.86).
population registers in five areas, (1) North Karelia Confirmatory factor analysis (with maximum likeli-
Province, (2) Kuopio Province, (3) South-Western hood estimation) was used to compare the applicability
Finland, (4) the cities of Helsinki and Vantaa in the of a one-factor and a correlated three-factor model of
capital area, and (5) the northern province of Oulu. the SOC items to the present data. An adequate fit
The sample was stratified according to gender, was obtained for both models after four error-term
10-year age groups, and area. In the age bracket covariances in the one-factor and one error-term
25e64 years there were 250 men and 250 women in covariance in the three-factor solution were released:
each 10-year age group in each region. In addition, c2 (61) ¼ 838.20, p < 0.001; comparison fit index
the sample included 500 men and 250 women aged (CFI) ¼ 0.95; root mean square error of approximation
65e74 from North Karelia Province and from the (RMSEA) ¼ 0.06 (CI 90%: 0.05e0.06); Akaike’s
H. Konttinen et al. / Social Science & Medicine 66 (2008) 2401e2412 2405

.54
.73 .46 .21
Mood 1 ME
.36 .56 .58 .31 .51 .64
.41 .64 .15
Pessimism .39
Worried Content Relaxed Upset Tense Calm 2 CO
.35 .49
.59 .60 .75 .76 .55 .71 .80 .42 .18
Failure 3 MA
.37 .61 .29
Unsatis- .54
faction 4 ME
.43 .66 .28
Guilty Anxiety .52
5 MA
.76
.26 -.65
Punish- .51 .60 .36
ment 6 CO
.40 .63 .27
Self hate Depression Sense of .52
7 ME
coherence
.30 .55 .73 .53
Self-
accusations 8 CO
.28 .52 .69 .47 .37
Self-punitive
wishes 9 CO
-.75
.24 .49 .20
.55 .30
Crying 10 MA
.07 .26
.37 .14
Irritability 11 CO
.25 .50
Social .68 .46
withdrawal 12 ME
.34 .59
Indeci- .71 .51
siveness 13 MA

Fig. 1. The first-order correlated three-factor model of the SOC (sense of coherence) scale, BDI-13 (Beck depression inventory without the eight
somatic symptoms) and STAI (state anxiety inventory) with standardized parameter estimates (note: ME ¼ meaningfulness item;
CO ¼ comprehensibility item; MA ¼ manageability item). The SOC items: 1 (ME) Do you have the feeling that you don’t really care what
goes on around you?; 2 (CO) Has it happened in the past that you were surprised by the behaviour of people whom you thought you knew
well?; 3 (MA) Has it happened that people whom you counted on disappointed you?; 4 (ME) Until now your life has had: no clear goals or
purpose at all e very clear goals and purpose; 5 (MA) Do you have the feeling that you’re being treated unfairly?; 6 (CO) Do you have the feeling
that you are in an unfamiliar situation and don’t know what to do?; 7 (ME) Doing the things you do every day is: a source of deep pleasure and
satisfaction e a source of pain and boredom; 8 (CO) Do you have very mixed-up feelings and ideas?; 9 (CO) Does it happen that you have feelings
inside you would rather not feel?; 10 (MA) Many people e even those with a strong character e sometimes feel like sad sacks (losers) in certain
situations. How often have you felt this way in the past?; 11 (CO) When something happened, have you generally found that: you overestimated or
underestimated its importance e you saw things in the right proportion; 12 (ME) How often do you have the feeling that there’s little meaning in
the things you do in your daily life?; 13 (MA) How often do you have feelings that you’re not sure you can keep under control?

information criterion (AIC) ¼ 898.20 for the one-factor Depressive symptoms


model, and c2 (61) ¼ 1006.32, p < 0.001; CFI ¼ 0.94; Depressive symptoms were measured on the Finnish
RMSEA ¼ 0.06 (CI 90%: 0.06e0.06); AIC ¼ 1066.32 version of the Beck depression inventory (BDI) (Beck,
for the three-factor model. The factors were highly Ward, Mendelson, Mock, & Erbaugh, 1961; Raitasalo,
correlated in the three-factor solution (rcoma ¼ 0.94; 1977). The BDI is a self-administered 21-item ques-
rmame ¼ 0.86; rcome ¼ 0.76). The three-factor model tionnaire covering the intensity of depressive symp-
could be considered slightly better than the one-factor toms. Each item includes four or five self-evaluative
model in which we had to release several error-term statements, which carry a numerical value from 1 to
covariances based on the largest modification indexes. 3 reflecting the intensity of the symptom. The respon-
However, we decided to use the one-factor solution in dents were asked to choose for each item one statement
the later confirmatory model because the correlations or more that best described their current situation. If
between the factors were so high (see also Klepp they had selected more than one option, which was
et al., 2007). rare, the one indicating the highest level of depression
2406 H. Konttinen et al. / Social Science & Medicine 66 (2008) 2401e2412

was coded on the data. We derived the BDI score by scores varied from 1 to 3, the higher ones indicating
summing up the ratings given to each of the 21 items. more symptoms.
It varied from 0 to 61 and showed good internal
consistency (Cronbach’s alpha ¼ 0.87). Smoking status
BDI covers eight somatic symptoms (distortion of Participants who reported that they had been
body image, work inhibition, sleep disturbance, smoking regularly (more than once a day) for at least
fatigability, loss of appetite, weight loss, somatic pre- a year, and had been doing so during the preceding
occupation and loss of libido) and 13 psychological month, were classified as current smokers, and were
symptoms. A measure of depressive symptoms includ- compared to former and never smokers in the analyses.
ing only these 13 psychological symptoms (referred to
as BDI-13 or cognitive depressive symptoms) was also Alcohol consumption
included as somatic symptoms could also indicate age- Alcohol consumption was assessed by means of
ing and physical health problems. a self-report questionnaire inquiring into the usual
quantity and frequency of consuming various bever-
Anxiety ages (Poikolainen, Poldkletnova, & Alho, 2002). The
The six-item short form of the state scale of the average weekly alcohol consumption (grams of pure
Spielberger State-Trait Anxiety Inventory (STAI) alcohol per week) was estimated based on this
(Marteau & Bekker, 1992) was used to measure state information.
anxiety: this has been found comparable to the full
form. The participants were asked to rate the six items Body mass index (BMI)
(e.g., ‘‘I am tense’’) from 1 (not at all) to 4 (very Body mass index was calculated as weight in
much) according to how they felt at that moment. kilograms (measured to the nearest 100 g in light
The ratings given to each item were summed so that clothing) divided by the square of height in meters
higher STAI scores reflected higher anxiety. The short (measured to the nearest 0.5 cm).
form of the STAI was internally consistent (Cronba-
ch’s a ¼ 0.85) and its range was from 6 to 24 in Blood pressure
this sample. Blood pressure was measured twice after 5 min of
rest in a sitting position and the average systolic value
Self-rated health was used in the analyses.
The question, What is your opinion about your cur-
rent health condition (1 ¼ very bad; 5 ¼ very good)? Cholesterol
was used to assess self-rated health. An enzymatic method was used to analyse the
serum total cholesterol value from serum samples.
Visits to a physician
Visits to a physician were covered by the question, Education
How many times have you visited a doctor during the Education was assessed on a continuous scale based
past year (12 months)? on the self-reported number of years of schooling.

Somatic and psychological symptoms Income


The participants were asked to indicate (from Income was measured in terms of the self-reported
1 ¼ often to 3 ¼ never) how often they had somatic household gross income during the previous 12 months,
or psychological symptoms in the previous 30 days. the response alternatives ranging from (1) less than
Somatic symptoms included tachycardia, trembling of 40,000 to (9) more than 320,000 Finnish marks (the
the hands, an irregular heart rate, dizziness, headache, Finnish currency is now the euro, one euro correspond-
and sweaty palms, while psychological symptoms in- ing to 5.95 Finnish marks).
cluded feeling confused when having to do something
quickly, feeling tense and nervous, having frightening Statistical methods
thoughts, feeling exhausted and overstrained, having
nightmares, depressiveness, and sleeplessness. The Pearson correlation coefficients were used in
mean scores of the ratings of the seven psychological assessing the associations between the SOC, depressive
and six somatic symptoms were calculated separately symptoms, anxiety and sociodemographic variables
(Cronbach’s a ¼ 0.81 and 0.67, respectively). The (years of education were age-adjusted), and the
H. Konttinen et al. / Social Science & Medicine 66 (2008) 2401e2412 2407

relationships between the SOC scale, BDI-13 and eight somatic symptoms from the BDI did not affect
STAI were further examined by means of confirmatory these correlations.
factor analyses (with maximum likelihood estimation). Confirmatory analysis was also used to investigate
Correlations with health and health-behaviour indica- the relations between SOC, cognitive depressive
tors were controlled for age. We used a method for symptoms and anxiety. A model was specified in
testing the significance of the difference between the which cognitive and affective BDI items loaded on
dependent correlation coefficients in order to deter- the depression factor, the SOC items on the SOC fac-
mine whether the SOC scale correlated with the health tor, and the STAI items on the anxiety factor. These
indicators and behaviours to a significantly different three latent variables were specified to correlate. This
degree than BDI, BDI-13 or STAI (Cohen & Cohen, model did not fit the data adequately based on the fit
1983). Finally, we tested equality of the strength of indexes: c2 (461) ¼ 6345.43, p < 0.001; SRMR ¼
the correlations among the SOC and various health- 0.04; CFI ¼ 0.88; RMSEA ¼ 0.06 (CI 90%: 0.05e0.06);
related variables between the lowest and the highest AIC ¼ 6479.43. After three error-term covariances
SOC tertiles (Cohen & Cohen, 1983). All the analyses between the SOC items were released on the basis of
except the factor analyses were conducted separately the largest modification indexes, the model provided
for men and women because this study included a satisfactory fit with the data: SRMR ¼ 0.04;
variables (depressive symptoms and health-related CFI ¼ 0.91; RMSEA ¼ 0.05 (CI 90%: 0.05e0.05);
measures) that normally feature gender differences. AIC ¼ 4712.18. The chi-square value was still signifi-
Cases with missing values were excluded. Amos 7 cant (c2 (458) ¼ 4572.18, p < 0.001), but the above-
and SPSS 15.0 were used. mentioned fit indexes suggested that this was due to
the large sample size. As Fig. 1 indicates, the depres-
Results sion factor correlated highly with both the SOC and
anxiety factors (r ¼ 0.75 and 0.76, respectively),
Table 1 presents the descriptive statistics for SOC, but it is possible to separate these constructs. The in-
depressive symptoms, anxiety and sociodemographic ter-correlation of the SOC and anxiety factors was
factors, and the Pearson correlations between these also strong (r ¼ 0.65).
variables separately for men and women. Correlation The associations of SOC, depressive symptoms
coefficients of less than 0.10 were not considered note- and anxiety with sociodemographic factors were
worthy, although they became statistically significant somewhat different (Table 1). Age was related to
due to the large sample size. depressive symptoms: the older men and women
The mean age was 51.3 years (SD ¼ 13.8) for the were more depressed. However, this association
2065 men and 47.1 years (SD ¼ 12.7) for the 2002 disappeared when the eight somatic symptoms of
women (Table 1). The SOC mean score was 65.6 the BDI were excluded. The younger participants
(SD ¼ 11.8) among men and 64.8 (SD ¼ 11.8) among showed more anxiety, but these correlations were
women, and was significantly higher for men (p ¼ weak (r ¼ 0.07 among males and r ¼ 0.06 among
0.006, h2 ¼ 0.002). The males (8.0, SD ¼ 7.3) had females). Years of education were age-adjusted
a significantly lower BDI mean score than females because of the rapid rise in the level of education
(8.7, SD ¼ 7.5) (p ¼ 0.001, h2 ¼ 0.003). A similar in recent decades (reducationage ¼ 0.45 for the men
phenomenon was observed when the eight somatic and reducationage ¼ 0.50 for the women, respec-
symptoms of the BDI were excluded. In contrast, the tively). Education was related to SOC, but not to
male and female STAI mean scores did not differ BDI, BDI-13 or STAI, after controlling for age:
(for the males: 10.4, SD ¼ 3.0; for the females: 10.4, more educated men and women had higher SOC
SD ¼ 3.1). scores (Table 1). In contrast, income was associated
As Table 1 shows, SOC, depressive symptoms and with SOC and depressive symptoms: SOC increased
anxiety were highly associated in both genders. The and depression decreased as the income level rose.
correlation between SOC and BDI was r ¼ 0.62 for We further investigated the discriminant validity of
all of the participants, and corresponding results were the SOC scale by comparing the associations of SOC,
found for BDI and STAI (r ¼ 0.64 in males and r ¼ depressive symptoms and anxiety with health indica-
0.65 in females). The association between SOC and tors and health behaviours. Higher SOC scores and
STAI (r ¼ 0.57 for men and r ¼ 0.53 for women) lower depression and anxiety scores were related to
was significantly lower than that between SOC and better self-rated health, fewer visits to a doctor, and
BDI in both genders (p < 0.01). The exclusion of the less somatic and psychological symptoms in both
2408 H. Konttinen et al. / Social Science & Medicine 66 (2008) 2401e2412

genders (Table 2). There were some differences be-

MineMax
16e91
0e46
0e30
6e24
25e74
1e40
tween the males and females: SOC, depressive symp-
toms and anxiety were all related to smoking and
alcohol consumption among the men, but no such asso-
ciations emerged among the women. However, SOC,

SOC ¼ sense of coherence scale; BDI ¼ Beck depression inventory; BDI-13 ¼ Beck depression inventory without the eight somatic symptoms; STAI ¼ state anxiety inventory.
Mean (SD)
(11.8)

(12.7)
(7.5)
(5.2)
(3.1)

(3.9)
BDI, BDI-13 and STAI showed no associations with
BMI, systolic blood pressure or total cholesterol levels
64.8

10.4
47.1
11.9
8.7
5.2 in either gender. These results suggest that SOC, de-
pressive symptoms and anxiety have inverse but other-
0.03* wise comparable relations with health indicators and
1.00
0.06

0.06
STAI

behaviours.
A comparison of the magnitude of the correlations
listed in Table 2 showed that BDI had significantly
BDI-13

0.03*

higher correlations with self-rated health, psychologi-


1.00
0.63

0.04
0.13

cal and somatic symptoms, and visits to a physician


than the SOC scale (p < 0.01). However, this differ-
ence remained significant only for psychological
1.00
0.95
0.65
0.14
0.06
0.16
Women (n ¼ 2002)
BDI

symptoms among males when the eight somatic


symptoms were excluded. Comparisons between the
SOC scale and STAI indicated that these variables
0.00*
1.00
0.62
0.62
0.53

0.11
0.17

were associated with all health indicators and health


SOC

behaviours to the same degree in both genders. Thus,


of the 54 comparisons of correlation coefficients, 45
The correlation coefficients between SOC, depressive symptoms, anxiety and the sociodemographic variables

indicated relationship equality.


MineMax

The SOC score tertiles were then used to investigate


19e91
0e47
0e34
6e24
25e74
1e54

whether the variation in a low or high SOC was more


strongly associated with health and health-behaviour
variables. As Table 3 shows, SOC was related to self-
Mean (SD)
(11.8)

(13.8)

rated health and visits to a physician among the men


(7.3)
(5.1)
(3.0)

(4.1)

and women whose SOC score fell into the lowest


65.6

10.4
51.3
11.9
8.0
4.6

SOC tertile, but there were no such relations in the high-


est tertile (p < 0.05). SOC was associated with somatic
and psychological symptoms in both the lowest and
0.01*
1.00
0.07

0.09
STAI

highest tertiles, but the correlations were stronger in


the lowest tertile among the men (p ¼ 0.035 for somatic
and p < 0.001 for psychological symptoms). The
BDI-13

strength of the associations among SOC and alcohol


0.03*
0.02*
1.00
0.65

0.13

consumption, smoking, BMI, blood pressure and cho-


lesterol were equal between the lowest and highest
tertiles, but all these correlations were low and most
1.00
0.94
0.64
0.13
0.04
0.16

of them were non-significant.


BDI
Men (n ¼ 2065)

*Not significant (at the p < 0.05 level).

Discussion
0.03*
1.00
0.62
0.63
0.57

0.12
0.20
SOC

We found strong relationships between the SOC


scale and measures of depressive symptoms and
Adjusted for age.

anxiety in a representative population sample of 25


Education (years)a

to 74-year-old Finnish people. These associations


became stronger when we applied structural equation
Age (years)

modelling, which takes measurement error into


BDI-13
Table 1

Income

account. The main aim of our study was to examine


STAI
SOC
BDI

the discriminant validity between the SOC, depression


H. Konttinen et al. / Social Science & Medicine 66 (2008) 2401e2412 2409

and anxiety scales by comparing their relations with

The strength of the correlation coefficients marked in bold is similar to the coefficients between SOC and different health-related variables. SOC ¼ sense of coherence scale; BDI ¼ Beck depression
81.0e225.0

Feeling confused when having to do something quickly, feeling tense and nervous, having frightening thoughts, feeling exhausted and overstrained, having nightmares, depressiveness, and
0.0e93.8
15.4e51.6

2.7e10.3
MineMax

1.0e3.0
1.0e3.0
0e92
health-related and sociodemographic variables. Educa-
1e5 tion was related only to SOC, but the associations of
SOC, cognitive depressive symptoms and anxiety
with health indicators and health behaviours were

(82.0)

(19.5)
Mean (SD)
(0.8)
(4.2)
(0.3)
(0.4)

(4.8)

(1.1)
practically identical in both genders. Our eventual
42.6
26.1
3.7
3.5
1.5
1.6

5.5
131.5
aim was to find out whether variation at the high end
of the SOC distribution is related to health measures.
The inter-correlations among the SOC and health
0.04*
0.01*

L0.01*
L0.28
0.17
0.33
0.54
0.06

L0.05
variables were stronger among the respondents in the
STAI

lowest than among those in the highest SOC tertile.


The SOC scores had a high negative association with
0.03*
0.04*
L0.04*
0.03*
BDI-13
L0.27
0.16
0.37
0.57
0.08

both depression (r ¼ 0.62 for the men and the women)


and anxiety (r ¼ 0.57 for the men and 0.53 for the
inventory; BDI-13 ¼ Beck depression inventory without the eight somatic symptoms; STAI ¼ state anxiety inventory; BMI ¼ body mass index.

women), which is consistent with previous findings


0.01*

L0.04*
0.02*
Women (n ¼ 1943)

0.34
0.21
0.41
0.60
0.09

0.07

(Eriksson & Lindström, 2005). Although factor analy-


BDI

ses suggested that it was possible to differentiate


between the SOC, cognitive depressive symptoms and
L0.02*

0.04*
0.26
L0.14
L0.32
L0.53
L0.08

L0.06

L0.05

anxiety variables, the estimated correlations were


SOC

even higher than those mentioned above. The relation-


ship between SOC and depressive symptoms was as
0.0e151.5

78.0e219.0
The age-adjusted correlation coefficients between SOC, depressive symptoms, anxiety and health-related variables

14.7e47.9
MineMax

strong as that between depressive symptoms and


1.0e2.8
1.0e3.0
0e99

2.4-10.5
1e5

anxiety. This was particularly interesting given


evidence from extensive studies on the close relation-
ship between depression and anxiety (Gotlib & Cane,
A rapid heart rate, shaking hands, an irregular heart rate, dizziness, headache, and sweaty hands.
(149.3)

(19.6)

1989).
Mean (SD)
(0.8)
(4.2)
(0.3)
(0.4)

(3.8)

(1.0)

A comparison of the SOC and the BDI items


26.9
3.5
2.7
1.4
1.5

5.6
103.2

140.3

revealed some content overlap. Four meaningfulness


items and one manageability item (numbered 1, 4, 7,
10 and 12, see Fig. 1) could be interpreted as reflecting
0.03*
0.02*
0.02*
L0.31
0.18
0.34
0.55
0.10
0.09
STAI

symptoms measured with the BDI (lack of satisfaction,


social withdrawal, sense of failure, guilty feeling and
self accusations), and two comprehensibility items
0.03*
L0.00*
0.04*
BDI-13
L0.29
0.18
0.34
0.56
0.14
0.13

(numbered 8 and 9, see Fig. 1) seemed to assess


negative emotional experiences in general. In addition,
two manageability items and one comprehensibility
L0.01*
0.02*
0.38
0.24
0.41
0.61
0.14
0.14
0.07

item (numbered 2, 3 and 5, see Fig. 1) were more likely


Men (n ¼ 1954)
BDI

to measure interpersonal trust/mistrust than the man-


Never/former smokers ¼ 1; current smokers ¼ 2.

ageability or comprehensibility dimensions of SOC.


L0.04*
0.02*
L0.02*
0.30
L0.13
L0.32
L0.51
L0.11
L0.11

In previous factor-analytic studies, these items have


SOC

formed their own factor, which has been interpreted


as reflecting interpersonal trust or mistrust (e.g.,
*Not significant (at the p < 0.05 level).
Visits to a physician (times/past year)

Alcohol consumption (abs. g/week)

Sandell et al., 1998). Factor analyses conducted on


Systolic blood pressure (mmHg)

the present data also indicated that the structure of


the SOC scale was rather ambiguous. This is no
Psychological symptomsb

wonder as half of the items could be interpreted as as-


Cholesterol (mmol/l)

sessing symptoms of depression or negative emotions.


Somatic symptomsa

The relationships of the SOC, depression and


Self-rated health

Smoking statusc

sleeplessness.

anxiety scales with years of education revealed some


BMI (kg/m2)

differences between these measures: educational level,


Table 2

controlled for age, was associated only with SOC.


a

c
b

Household annual income, however, was associated


2410 H. Konttinen et al. / Social Science & Medicine 66 (2008) 2401e2412

Table 3
The correlation coefficients between SOC and health-related variables by SOC tertiles
Men Women
Low SOCa Medium SOCb High SOCc Low SOCa Medium SOCb High SOCc
(n ¼ 626) (n ¼ 717) (n ¼ 741) (n ¼ 708) (n ¼ 682) (n ¼ 678)
Self-rated health 0.18*** 0.10** 0.06 0.20*** 0.01 0.00
Visits to a physician (times/past year) 0.13** 0.01 0.01 0.17*** 0.05 0.04
Somatic symptoms 0.25*** 0.06 0.14*** L0.21*** 0.01 L0.13**
Psychological symptoms 0.39*** 0.10* 0.20*** L0.37*** 0.05 L0.31***
Alcohol consumption (abs. g/week) L0.10* 0.04 L0.03 L0.02 0.03 L0.05
Smoking statusd L0.06 0.05 L0.02 L0.06 0.05 L0.09*
BMI (kg/m2) L0.01 0.03 0.01 L0.05 0.02 0.05
Systolic blood pressure (mmHg) 0.08* 0.00 0.04 L0.02 0.03 0.08*
Cholesterol (mmol/l) 0.04 0.02 L0.04 L0.03 0.01 0.03
***p < 0.001; **p < 0.01; *p < 0.05.
The correlation coefficients marked in bold are equal in strength between the lowest and highest tertiles. SOC ¼ sense of coherence scale; BMI ¼
body mass index.
a
Scores 16e60.
b
Scores 61e71.
c
Scores 72e91.
d
Never/former smokers ¼ 1; current smokers ¼ 2.

with both SOC and depression. In addition, the levels sizeable, becoming statistically significant due to the
of both SOC and depressive symptoms varied accord- large sample size (rSOCpsychological symptoms ¼ 0.51
ing to gender: women had a significantly lower SOC and rBDI-13psychological symptoms ¼ 0.56). Furthermore,
mean score and a higher BDI mean score than men. the strength of the correlations of the SOC scale,
Age was not related to SOC in these data, although BDI, BDI-13 and STAI with health behaviours
not all population studies have found SOC to be (smoking and alcohol consumption) was consistent.
independent of age (Larsson & Kallenberg, 1996). We also examined whether variation at the high end
Antonovsky (1987) developed SOC as an explana- of the SOC was related to health measures, as depres-
tory concept related to health. Hence, the main interest sion inventories do not extend to the positive end. The
in this study was to compare the associations of SOC, resulting analyses with the SOC tertiles indicated that
depressive symptoms and anxiety with health the variation at the low end was more strongly associ-
indicators and behaviours. Consistent with the results ated with health variables than the variation at the high
of previous studies (Eriksson & Lindström, 2006; end. This is consistent with the findings reported by
Friedman & Booth-Kewley, 1987), SOC was positively Kivimäki et al. (2000): it was only a low and not
and depressive symptoms and anxiety were negatively a high SOC that predicted health (sickness absence).
associated with various health-related measures. The These results could be interpreted to imply that a strong
full 21-item BDI includes eight somatic symptoms SOC is not necessary for good health, although it may
that, in addition to depression, are directly related to have a positive impact on other aspects of the quality
ageing and physical health (rageBDI ¼ 0.13 for the of life.
men and 0.14 for the women). Thus, to avoid possible The present study included quite a wide range of
bias, the 13-item BDI, which includes only cognitive health-related measures: self-reported health indicators
and affective symptoms (rageBDI-13 ¼ 0.03 for the and health behaviours, and clinically measured health
men and 0.03 for the women), is more appropriate in indicators (BMI, systolic blood pressure and total
terms of comparing the relationships of SOC and de- cholesterol). SOC, depressive symptoms and anxiety
pressive symptoms with health indicators. The SOC all had the strongest correlations with self-assessed
scale, BDI-13 and STAI were associated to an equal health variables. It is possible that self-reported health
magnitude with all the subjective and objective health indicators are confounded by psychological character-
indicators used in the present study. The only exception istics and emotional states. For example, people with
was that BDI-13 had a significantly higher correlation depressive symptoms or anxiety may report more phys-
with psychological symptoms among the men than the ical symptoms, and may visit a physician more often
SOC measure. However, this difference was not because negative mood states result in negative biases
H. Konttinen et al. / Social Science & Medicine 66 (2008) 2401e2412 2411

in the evaluation and categorisation of stimuli (Cohen Antonovsky, A. (1979). Health, stress and coping. San Francisco:
& Rodriguez, 1995). In this study, SOC, depression Jossey-Bass.
Antonovsky, A. (1987). Unraveling the mystery of health: How
and anxiety were not related to clinically measured people manage stress and stay well. San Francisco: Jossey-Bass.
health variables. Antonovsky, A. (1993). The structure and properties of the sense of
To our knowledge, this is the first study to compare coherence scale. Social Science & Medicine, 36(6), 725e733.
the relationships of SOC, depressive symptoms and Bandura, A. (1977). Self-efficacy: toward a unifying theory of
anxiety with health-related variables in a representative behavioral change. Psychological Review, 84, 191e215.
Beck, A. T. (1967). Depression: Causes and treatment. Philadelphia:
population sample. A few earlier studies have compared University of Pennsylvania Press.
the associations of SOC and other personality traits with Beck, A. T. (1987). Cognitive models of depression. Journal of
health variables. The SOC scale has shown stronger cor- Cognitive Psychotherapy, 1(1), 5e37.
relations with different health and well-being measures Beck, A. T., Ward, C. H., Mendelson, M., Mock, J., & Erbaugh, J.
than the self-efficacy, hardiness, locus of control, self- (1961). An inventory for measuring depression. Archives of
General Psychiatry, 4, 561e571.
esteem and optimism scales, for example (Pallant & Breslin, F. C., Hepburn, C. G., Ibrahim, S., & Cole, D. (2006).
Lae, 2002; Smith & Meyers, 1997). Similar associations Understanding stability and change in psychological distress
found between the SOC, depression and anxiety and sense of coherence. Journal of Applied Social Psychology,
measures suggest that SOC has more in common with 36(1), 1e21.
them than other positive psychology concepts. Clark, L. A., & Watson, D. (1991). Tripartite model of anxiety and
depression: psychometric evidence and taxonomic implications.
In the present study, we used the 13-item SOC Journal of Abnormal Psychology, 100(3), 316e336.
questionnaire, and it is possible that the heterogeneity Cohen, J., & Cohen, P. (1983). Applied multiple regression/correla-
of the SOC construct would be better captured in the tion analysis for the behavioral sciences. Hillsdale: Erlbaum.
full 29-item questionnaire. Furthermore, the depression Cohen, S., & Rodriguez, M. S. (1995). Pathways linking affective
and anxiety inventories we used are not unproblematic disturbances and physical disorders. Health Psychology, 14(5),
374e380.
either: it is possible that one common factor, such as Eriksson, M., & Lindström, B. (2005). Validity of Antonovsky’s
negative emotionality, explains the strong and inverse sense of coherence scale: a systematic review. Journal of
relationship of the SOC with depression and anxiety Epidemiology & Community Health, 59(6), 460e466.
(Frenz et al., 1993). Indeed, the SOC questionnaire Eriksson, M., & Lindström, B. (2006). Antonovsky’s sense of
might measure negative emotionality, and according coherence scale and the relation with health: a systematic
review. Journal of Epidemiology & Community Health, 60,
to the tripartite model of depression and anxiety (Clark 376e381.
& Watson, 1991), negative emotions common to both Feldt, T., Lintula, H., Suominen, S., Koskenvuo, M., Vahtera, J., &
depression and anxiety are mainly responsible for the Kivimäki, M. (2006). Structural validity and temporal stability
strong association between the measures of these two of the 13-item sense of coherence scale: prospective evidence
constructs. from the population-based HeSSup study. Quality of Life
Research, 16, 483e493.
In conclusion, our study provides additional Feldt, T., Metsäpelto, R.-L., Kinnunen, U., & Pulkkinen, L. (2007).
evidence of the overlap between the SOC scale and Sense of coherence and five-factor approach to personality.
measures of depression and anxiety. This is consistent European Psychologist, 12(3), 165e172.
with the findings reported by Breslin et al. (2006), Flannery, R. B., & Flannery, G. J. (1990). Sense of coherence,
Gruszczyńska (2006) and Korotkov (1993). It seems life stress, and psychological distress: a prospective
methodological inquiry. Journal of Clinical Psychology,
that current psychometric methods are able to separate 46(4), 415e420.
highly correlated measures, and these small differences Frenz, A. W., Carey, M. P., & Jorgensen, R. S. (1993). Psychometric
could have value in some research settings. However, evaluation of Antonovsky’s sense of coherence scale. Psycholog-
in view of our almost equal associations with health ical Assessment, 5(2), 145e153.
indicators and behaviours, it would be useful to bring Friedman, H. S., & Booth-Kewley, S. (1987). The ‘disease prone
personality’: a meta-analytic view of the construct. American
these concepts into the same scientific discussion. Psychologist, 42(6), 539e555.
This kind of research could have a positive impact in Geyer, S. (1997). Some conceptual considerations on the sense of
the application of these constructs to theory-based coherence. Social Science & Medicine, 44(12), 1771e1779.
health-improvement programmes. Gotlib, I. H., & Cane, D. B. (1989). Self-report assessment of
depression and anxiety. In: P. C. Kendall, & D. Watson (Eds.),
Anxiety and depression: Distinctive and overlapping features
References (pp. 131e169). San Diego: Academic Press.
Gruszczyńska, E. (2006). What is measured by the orientation to life
Abramson, L. Y., Metalsky, G. I., & Alloy, L. B. (1989). Hopeless- questionnaire? Construct validity of the instrument for the sense
ness depression: a theory-based subtype of depression. of coherence measurement. Polish Psychological Bulletin, 37(2),
Psychological Review, 96(2), 358e372. 74e83.
2412 H. Konttinen et al. / Social Science & Medicine 66 (2008) 2401e2412

Kaplan, G. A. (1995). Where do shared pathways lead? Some commonly used laboratory markers. Alcohol & Alcoholism,
reflections on a research agenda. Psychosomatic Medicine, 57, 37(6), 573e576.
208e212. Raitasalo, R. (1977). Depression and its connections with need for
Kivimäki, M., Feldt, T., Vahtera, J., & Nurmi, J. (2000). Sense of psychotherapy. Helsinki: Research Institute for Social Security.
coherence and health: evidence from two cross-lagged longitudi- Rotter, J. B. (1966). Generalized expectancies for internal versus ex-
nal samples. Social Science & Medicine, 50, 583e597. ternal control of reinforcement. Psychological Monographs, 80(1).
Klepp, O. M., Mastekaasa, A., Sorensen, T., Sandanger, I., & Sandell, R., Blomberg, J., & Lazar, A. (1998). The factor structure of
Kleiner, R. (2007). International Journal of Methods in Psychiat- Antonovsky’s sense of coherence scale in Swedish clinical and
ric Research, 16(1), 11e22. nonclinical samples. Personality and Individual Differences,
Kobasa, S. C. (1982). The hardy personality: toward a social psychol- 24(5), 701e711.
ogy of stress and health. In: J. Suls, & G. Sanders (Eds.), Social Smith, T. L., & Meyers, L. S. (1997). The sense of coherence: its
psychology of health and illness (pp. 3e33). Hillsdale: Erlbaum. relationship to personality, stress and health measures. Journal
Korotkov, D. (1993). An assessment of the (short-form) sense of of Social Behavior and Personality, 12(2), 513e526.
coherence personality measure: issues of validity and well-being. Strümpfer, D. J. W., Gouws, J. F., & Viviers, M. R. (1998). Antonov-
Personality and Individual Differences, 14, 575e583. sky’s sense of coherence scale related to negative and positive
Kravetz, S., Drory, Y., & Florian, V. (1993). Hardiness and sense of affectivity. European Journal of Personality, 12(6), 457e480.
coherence and their relation to negative affect. European Journal Suls, J., & Bunde, J. (2005). Anger, anxiety, and depression as risk
of Personality, 7(4), 233e244. factors for cardiovascular disease: the problems and implications
Kuuppelomäki, M., & Utriainen, P. (2003). A 3 year follow-up study of overlapping affective dispositions. Psychological Bulletin,
of health care students’ sense of coherence and related smoking, 131(2), 260e300.
drinking and physical exercise factors. International Journal of Suominen, S., Blomberg, H., Helenius, H., & Koskenvuo, M. (1999).
Nursing Studies, 40(4), 383e388. Sense of coherence and health e does the association depend on
Larsson, G., & Kallenberg, K. (1996). Sense of coherence, socioeco- resistance resources? A study of 3115 adults in Finland. Psychol-
nomic conditions and health. European Journal of Public Health, ogy & Health, 14(5), 937e948.
6(3), 175e180. Suominen, S., Helenius, H., Blomberg, H., Uutela, A., &
Lundberg, O. (1997). Childhood conditions, sense of coherence, Koskenvuo, M. (2001). Sense of coherence as a predictor of
social class and adult ill health: exploring their theoretical and subjective state of health: results of 4 years of follow-up of adults.
empirical relations. Social Science & Medicine, 44(6), 821e831. Journal of Psychosomatic Research, 50(2), 77e86.
Lundberg, O., & Nyström Peck, M. (1994). Sense of coherence, Surtees, P., Wainwright, N., Luben, R., Khaw, K. T., & Day, N.
social structure and health. Evidence from a population survey (2003). Sense of coherence and mortality in men and women in
in Sweden. European Journal of Public Health, 4, 252e257. the EPIC-Norfolk United Kingdom prospective cohort study.
Marteau, T. M., & Bekker, H. (1992). The development of a six-item American Journal of Epidemiology, 158(12), 1202e1209.
short-form of the state scale of the Spielberger State-Trait Vartiainen, E., Jousilahti, P., Alfthan, G., Sundvall, J., Pietinen, P., &
Anxiety Inventory (STAI). British Journal of Clinical Psychol- Puska, P. (2000). Cardiovascular risk factor changes Finland,
ogy, 31, 301e306. 1972e1997. International Journal of Epidemiology, 29, 49e56.
Midanik, L. T., Soghikian, K., Ransom, L. J., & Polen, M. R. (1992). Veenstra, M., Moum, T., & Roysamb, E. (2005). Relationships
Alcohol problems and sense of coherence among older adults. between health domains and sense of coherence: a two-year
Social Science & Medicine, 34, 43e48. cross-lagged study in patients with chronic illness. Quality of
Mineka, S., Watson, D., & Clark, L. A. (1998). Comorbidity of Life Research, 14, 1455e1465.
anxiety and unipolar mood disorders. Annual Review of Psychol- Watson, D., & Clark, L. A. (1984). Negative affectivity: the disposi-
ogy, 49, 377e412. tion to experience aversive emotional states. Psychological
Pallant, J. F., & Lae, L. (2002). Sense of coherence, well-being, cop- Bulletin, 96(3), 465e490.
ing and personality: further evaluation of the sense of coherence Watson, D., & Kendall, P. C. (1989). Understanding anxiety and
scale. Personality and Individual Differences, 33(1), 39e48. depression: their relation to negative and positive affective states.
Poikolainen, K., Poldkletnova, I., & Alho, H. (2002). Accuracy of In: P. C. Kendall, & D. Watson (Eds.), Anxiety and depression:
quantity-frequency and graduated frequency questionnaires in Distinctive and overlapping features (pp. 3e26). San Diego:
measuring alcohol intake: comparison with daily diary and Academic Press.

Potrebbero piacerti anche