Sociodemographic, disease status, and illness perceptions predictors of global self-ratings of health and quality of life among those with coronary heart disease--one year follow-up study.

Abstract

This one-year follow-up study (n = 130 at baseline, n =2745 at follow-up, aged 45-74 years) examined the relationship of patients' perceptions of coronary heart disease (CHD) and illness-related factors with global health status and global quality of life (QOL) ratings. The independent variables were CHD history (myocardial infarction, revascularisation), CHD severity (use of nitrates, CHD risk factors and co-morbidities) and illness perceptions. In multivariate regression analysis, CHD history and severity explained 13% of variance in global health status and 8% in global QOL ratings at the baseline. Illness perceptions increased the share of explained variance by 18% and 16% respectively. In the follow-up, illness perceptions explained a significant but modest share of variance in change in health status and QOL when baseline health status and QOL and CHD severity were adjusted for more symptoms being attributed to CHD, severe perceived consequences of CHD, as well as a weak belief in the controllability of CHD were related to poor global health status and QOL ratings. In structural path models associations of CHD severity factors were mediated by illness perceptions. The association of disease severity with dependent variables was weaker after controlling for illness perceptions. Cognitive representations of CHD contribute to both global health status and QOL ratings and they also mediate the associations between CHD severity and well-being. No gender differences were found in associations of illness perceptions with health status or QOL ratings.

Full text

Page 1

Aalto, A.M., Aro, A.R., Weinman, J., Heijmans, M., Manderbacka, K., Elovainio, M.
Sociodemografhic, disease status, and illness perceptions predictors of global self-ratings of
health and quality of life among those with coronary heart disease: one year follow-up study.
Quality of Life Research: 15, 2006, nr. 8, p. 1307-1322

Postprint Version 1.0
Journal website http://www.springerlink.com
Pubmed link http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dop
t=AbstractPlus&list_uids=16826444&query_hl=20&itool=pubmed_docsum
DOI 10.1007/s11136-006-0010-3

This is a NIVEL certified Post Print, more info at http://www.nivel.eu -1-
Sociodemographic, disease status, and illness perceptions
predictors of global self-ratings of health and quality of life
among those with coronary heart disease – one year follow-
up study
ANNA-MARI AALTO1, ARJA R. ARO2,3, JOHN WEINMAN4, MONIQUE HEIJMANS5, KRISTIINA MANDERBACKA1
& MARKO ELOVAINIO1
1 Health services research, STAKES (National research and development centre for welfare and health),
Lintulahdenkuja 4, Helsinki, Finn-00531, Finland (E-mail: anna-mari.aalto@stakes.fi);
2 University of Southern Denmark, Esbjerg, Denmark;
3 Erasmus Medical Center, Rotterdam, The Netherlands;
4 Unit of Psychology, United Medical and Dental School of Guy’s and St Thomas’s Hospitals, KCL, London,
UK;
5 NIVEL, Utrecht, The Netherlands

ABSTRACT
This one-year follow-up study (n = 3130 at baseline, n = 2745 at follow-up, aged 45–74 years)
examined the relationship of patients’ perceptions of coronary heart disease (CHD) and illness-
related factors with global health status and global quality of life (QOL) ratings. The
independent variables were CHD history (myocardial infarction, revascularisation), CHD
severity (use of nitrates, CHD risk factors and co-morbidities) and illness perceptions. In
multivariate regression analysis, CHD history and severity explained 13% of variance in global
health status and 8% in global QOL ratings at the baseline. Illness perceptions increased the
share of explained variance by 18% and 16% respectively. In the follow-up, illness perceptions
explained a significant but modest share of variance in change in health status and QOL when
baseline health status and QOL and CHD severity were adjusted for more symptoms being
attributed to CHD, severe perceived consequences of CHD, as well as a weak belief in the
controllability of CHD were related to poor global health status and QOL ratings. In structural
path models associations of CHD severity factors were mediated by illness perceptions. The
association of disease severity with dependent variables was weaker after controlling for illness
perceptions. Cognitive representations of CHD contribute to both global health status and QOL
ratings and they also mediate the associations between CHD severity and well-being. No gender
differences were found in associations of illness perceptions with health status or QOL ratings.
INTRODUCTION
Coronary Heart Disease (CHD) is one of the most common causes of premature death among both men and
women in western societies. In chronic illnesses, which often involve complicated treatment regimens with
lifestyle adjustments and daily medication, one of the challenges is to find a balance between the demands
and restrictions posed by the illness and it’s treatment on one hand and the various challenges and demands

Page 2

Aalto, A.M., Aro, A.R., Weinman, J., Heijmans, M., Manderbacka, K., Elovainio, M.
Sociodemografhic, disease status, and illness perceptions predictors of global self-ratings of
health and quality of life among those with coronary heart disease: one year follow-up study.
Quality of Life Research: 15, 2006, nr. 8, p. 1307-1322

of every day life on the other hand. For example following lifestyle restriction may be difficult to adjust with
the demands of the working life. In recent years, the outcomes of illness processes have been increasingly
assessed from the patients’ perspective in terms of quality of life (QOL). QOL is a broad multidimensional
concept in which the focus is on the individual’s subjective experience. Domains such as emotional health,
physical health, social networks, material resources and work or productive activity have been proposed as
components of QOL. According to Leventhal and Colman [1], QOL judgement is a process in which the
individual assesses his or her personal experience in various life-domains and then integrates these into one
overall judgement.
CHD affects various domains of QOL [2, 3] and the effects are more severe in younger patients [3, 4]. The
major complication of CHD, myocardial infarction (MI), undermines QOL and functioning [5, 6], causes
uncertainty and worry [7] and requires readjustment from the patient and his or her social environment [8].
The survival rates of MI have increased due to more effective treatments and secondary prevention, and
therefore more patients will live their lives with the consequences of MI. Another common manifestation of
CHD, angina pectoris (AP), also has negative impact on QOL in terms of both physical functioning and
emotional health [9, 10]. Presence of common epidemiological risk factors of CHD, such as diabetes [11]
may further impair QOL among persons with CHD. Treatment of CHD by revascularisation (coronary artery
bypass grafting (CABG) or percutaneous transluminal coronary angioplasty (PTCA)), however, clearly has
beneficial QOL effects: Brorson et al. [12] reported improvements in physical functioning, pain, quality of
sleep, emotional health and general health perceptions after CABG and PTCA.
CHD has typically been considered to be a male disease [13] even though it is a major cause of death and
disability among women as well. Gender differences have been reported in the clinical manifestation of
CHD, women presenting more often with AP while MI is more common among men [14]. Furthermore,
gender inequalities have been reported in the treatment of CHD with women receiving less CHD-related
surgical operations than men [15]. Women also seem to be sensitive to different kind of risk factors
compared to men [16], with psychosocial factors, such as work climate or burnout, being of more importance
among women. Furthermore, women seem to adjust and cope less effectively than men after MI since they
tend to report more impaired QOL and more psychological morbidity (see Brezinka and Kittel [17] for
review).
From the point of view of the counselling and rehabilitation of patients with CHD, knowledge about the
processes which facilitate adjustment and QOL in living with heart disease is important. Individual
differences in adjusting to a chronic illness can’t solely be explained by disease severity, psychosocial
processes and recourses are important too. According to the Leventhal et al.’s Self- Regulation Model (SRM)
[18], people develop personal representations about the nature of the illness to understand the health threats
and use these to regulate both their behavioural and emotional reactions to the illness. Five central
components of these implicit theories are (1) the experienced symptoms which are attributed to the illness
(identity), (2) beliefs about causes of illness (cause), (3) beliefs about the curability or controllability of the
illness (cure/control), (4) perceived consequences of the illness in the person’s everyday life (consequences),
(5) expectancies about duration of the illness (time-line). These representations of illness affect the salience,
importance and meaning of the domains involved in making the QOL judgements (1).
In CHD, illness perceptions have mostly been studied as predictors of rehabilitation attendance and return
to work. Better attendance to rehabilitation has been related to a stronger belief in the controllability of heart
disease [19–21], less perceived symptoms associated with heart disease [21], and attribution of own heart
disease to life style factors [20]. A slower return to work and resumption of social and domestic duties has
been related to more negative perceived consequences of MI [19]. Figueiras and Weinman [22] reported that
similarity between patient’s and spouse’s illness perceptions was related to recovery from a heart attack and
changes in health behaviours.
The aim of the present study is to examine how illness perceptions together with disease severity and
history and sociodemographic factors are related to self-rated global health status and global QOL among
persons with CHD, as well as to examine whether illness perceptions mediate the relationships between
CHD-related factors and global health status and QOL self-ratings. Due to the reported gender differences in
presentation and outcomes of CHD, we were also interested in whether these associations differed by gender.
This is a NIVEL certified Post Print, more info at http://www.nivel.eu -2-

Page 3

Aalto, A.M., Aro, A.R., Weinman, J., Heijmans, M., Manderbacka, K., Elovainio, M.
Sociodemografhic, disease status, and illness perceptions predictors of global self-ratings of
health and quality of life among those with coronary heart disease: one year follow-up study.
Quality of Life Research: 15, 2006, nr. 8, p. 1307-1322

METHODS
The sample
In Finland all those fulfilling certain diagnostic criteria of CHD are entitled to elevated reimbursement for
the costs of medication for CHD. The eligibility for the special reimbursement right requires a medical
certificate by the attending doctor, usually an internist or a cardiologist, on following criteria: (1) diagnosed
chronic angina pectoris symptoms responding to nitrates in the presence of unequivocal ECG changes (on
exercise or at rest) or (2) diagnosed acute myocardial infarction, or (3) revascularisation operation or (4)
coronary heart disease diagnosed in angiography. The certificates are reviewed and approved by a specialist
physician at the Social Insurance Institution which also maintains a register of all those who are eligible.
From this register 5009 Finnish-speaking people, aged 45–74 years in January 2001 and who had received
the right to medication reimbursement due to CHD (nitrates or secondary prevention medicines, such as
statines) during 1994–1995, were drawn for the study sample. In age groups younger than this the prevalence
of CHD had been very low. Older persons might have had difficulties to understand and to fill in the
questionnaire, therefore to guarantee a reasonable response rate we used 74 as the upper age limit. To
guarantee a sufficient number of women in the sample, stratification by gender (50% each) was followed by
random sampling within each gender group. Altogether 3539 questionnaires were returned, of which 409
were rejected due to incomplete data (missing values exceeding 20%), leaving 3130 people in the study
sample at baseline (62.5% of the original sample). Seven persons had died between the day the sample was
drawn and mailing of the questionnaire (mortality register information). We received approximately 200
calls from the persons who got the questionnaire but were not willing to participate, and the most common
reason mentioned for non-participation was poor health status or hospitalization. Yet in some cases the
person was just un-willing to participate and give consent to use his/her information for study purposes. The
follow-up questionnaire was sent to respondents (n = 3082, 48 persons died during the follow-up period) one
year later. A total of 2650 questionnaires were returned, giving a response rate of 84% (53% of the original
sample).
For analysis of attrition, census based data on gender, age, and education was individually linked to the
original sample (5009) in Statistics Finland using the unique personal identification codes, which were
removed from the data before handing them to the research team. This allowed us to directly compare the
respondents and non-respondents in terms of age, education and gender. Non-respondents were (p<0.001)
older, more frequently women and less well educated than the participants. The non-respondents in follow-
up survey differed from respondents in being less educated (p<0.001). The participants were asked for a
written consent to combine their questionnaire data with information about their use of hospital services
derived from the Hospital Discharge Register; 92% of respondents consented. The study protocol was
approved by the Research Ethics Committee of the National Research and Development Centre for Welfare
and Health.
Measures
This is a NIVEL certified Post Print, more info at http://www.nivel.eu -3-
Dependent variables
Global health status rating was measured by the European standard single item question for self-rated health
status: ‘‘Is your health at present?’’ (1) very good, (2) good, (3) moderate, (4) poor, (5) very poor. The
European version was chosen because it was expected to differentiate more effectively in a sample of
patients with a chronic illness than the US version, which has more positive response alternatives and only
one alternative for poor health status. For the analysis, the coding was reversed so that higher scores
indicated good health. QOL was measured by a visual analogue scale (VAS). Respondents were asked to rate
their overall quality of life on a 100 mm scale (0–100), where 0 indicates worst possible QOL and 100 best
possible QOL. Validity and reliability have been reported for both global measures. Content reliability of
self-rated global health status is supported by it’s close associations with chronic conditions and their
functional consequences [23] while evidence of predictive validity is given by studies showing that self-rated
global health status is an independent predictor of mortality even when other risk factors are accounted for
[24]. Test–retest reliability has also been reported for it the kappa coefficient being 0.70 [25], which
indicates good reliability [26]. For global self-rating of QOL, Boer et al. [27] reported validity in terms of it’s
associations with established health-related QOL measures (MOS-20 and Rotterdam Symptoms Check List)
and good test–re-test reliability (r = 0.87). Global health status and global QOL are distinct constructs, the

Page 4

Aalto, A.M., Aro, A.R., Weinman, J., Heijmans, M., Manderbacka, K., Elovainio, M.
Sociodemografhic, disease status, and illness perceptions predictors of global self-ratings of
health and quality of life among those with coronary heart disease: one year follow-up study.
Quality of Life Research: 15, 2006, nr. 8, p. 1307-1322

physical dimensions of health being more important for global self-rated health status and emotional
dimensions for global QOL rating [28, 29].
This is a NIVEL certified Post Print, more info at http://www.nivel.eu -4-
Independent variables
Sociodemographic background was measured by age, gender (0 = female, 1 = male) and education. In
Finland every permanent resident is given a personal identity number, which is composed of the date of birth
and a personal identification code which is an odd number for men and even number for women. Information
about age and gender was based on this identity number. Information about education was based on the self-
reported level of basic education and of vocational training. To provide a measure of total education, the
levels given in the two responses were converted into years and then summed to give a total observed score
ranging from 8 to 17 years.
CHD-related factors: Information on previous myocardial infarction (MI) (0 = no, 1 = yes) and coronary
revascularisation (CABG or PTCA, 0 = no, 1 = yes) was derived from the Hospital Discharge Register for
those who had given their consent to combine register-based data with their self-report data (92%). For those
who refused their consent, self-report information regarding the occurrence of MI or revascularisation was
used. Since those undergoing PTCA may have milder diseases status which could potentially complicate the
interpretation of the results, we did a preliminary analysis using PTCA and CABG as distinct variables.
However, the association of these variables with dependent variables were in the same direction and using
them separately did not lead to any additional increases in explained variance. To achieve more power and a
simpler final model we combined these variables for the analysis.
Since we did not have access to physiological indicators of CHD severity, we therefore utilized the use of
short-term nitrates at least once a week as a proxy measure of disease severity (5 = daily, 4 = few times a
week, 3 = few times a month, 2 = less often, 1 = never). Short-term nitrates are prescribed for symptoms of
angina and can therefore be seen as an indicator of the frequency of angina. In addition, the respondents were
asked about the presence of CHD-related co-morbidities (blood pressure, cardiac failure, high serum
cholesterol, arrhythmia, diabetes, ischaemic attack). Two variables were constructed for additional indicators
of disease severity. Heart failure or arrythmia may be seen as consequences of CHD, therefore if either of
these was mentioned the respondent was considered to have CHD-related co-morbidities (1 = yes, 0 = no).
More cautious term co-morbidity is used here instead of complications, since it is possible that these two
conditions exist as independent conditions unrelated to CHD. Presence of physiological risk factors was
coded (1 = yes, 0 = no) if the respondent mentioned having high blood pressure, high cholesterol, or diabetes
or being a current smoker.
Illness perceptions were assessed by the Illness Perceptions Questionnaire [30]. The scales used in the
present study were illness identity (12 items, alpha coefficient α = 0.83) cure/control (6 items, α = 0.52),
consequences (5 items, α = 0.69) and time-line (3 items, α = 0.70). The alpha coefficient of the cure/control
scale was relatively low, as has also been found in other studies [31]. Since excluding any subset of items did
not significantly improve the alpha coefficient and the preliminary analysis showed that the six-item scale
produced the most predictive correlations, all the items were retained in this scale in spite of the low internal
consistency.
For the illness identity items, the response scale ranged from 0 (= never) to 3 (= all the time). On other
items, the response scale ranged from 1 (strongly disagree) to 5 (strongly agree). High scores on these IPQ-
scales indicate a higher number of symptoms attributed to CHD (identity), strong belief in the controllability
of CHD (cure/ control), severe perceived consequences of CHD and long expected duration of CHD (time-
line).
To measure the ‘‘causes’’ dimension of illness perceptions, we used a scale adapted from the IPQ-R [32]
supplemented by 8 additional items related to CHD risk. Based on preliminary explorative factor analysis
[33] and a further confirmatory factor analysis for the present study in which items loading strongly on
several factors were excluded (see Appendix) three separate sum scores were computed for the present study:
attribution to stress factors (4 items, alpha = 0.76), CHD risk factors (5 items alpha = 0.72) and internal
factors (3 items, alpha = 0.70).
Statistical analysis
Univariate baseline associations of global health status and QOL ratings with independent variables were
examined by Pearson correlation coefficients in continuous variables and by an analysis of variance in
categorical variables. The non-linearity of the relationship between age and global health status and QOL

Page 5

Aalto, A.M., Aro, A.R., Weinman, J., Heijmans, M., Manderbacka, K., Elovainio, M.
Sociodemografhic, disease status, and illness perceptions predictors of global self-ratings of
health and quality of life among those with coronary heart disease: one year follow-up study.
Quality of Life Research: 15, 2006, nr. 8, p. 1307-1322

was examined by regressing the dependent variables for age and square of age simultaneously. When the
‘‘square of age’’ term was statistically significant, the relationship of age and dependent variables was
examined in more detail by comparing global health status and QOL in the 5-year age groups.
Series of multivariate regression analyses were used to examine the independent associations of
sociodemographic factors, CHD-related factors, and illness perceptions with global self-rated health status or
QOL. Analyzing all the individual gender interactions in relationships between independent and dependent
variables would have yielded 36 interaction terms and the possibility of significant interaction terms
produced by chance would have been high, particularly in the absence of pre-formulated hypotheses about
the nature of the interactions. Therefore, to assess gender interactions, the Chow test statistics1 was
computed [34] in the multivariate analysis. Chow test is commonly used to test the structural changes in
model parameters and it can be used to examine if the parameters of the model are the same in different
time-points or in different sub samples [35].
Mediation effect of CHD illness perceptions and attributions in the relationship between CHD severity and
global health status and QOL self-ratings was examined first in multivariate regression analyses. We applied
the rules of Baron and Kenny [36], according to whom to show that a variable mediates the relationship
between two other variables, all three variables must correlate significantly with each other and the
relationship between the independent variable (CHD severity indicators) and dependent variable (global
health status and QOL self-ratings) should be reduced when the effect of the mediator (illness perceptions) is
adjusted for. In partial mediation the independent variable may show a significant but lower main effect
when the moderator is adjusted for, in complete mediation there is not main effect of the independent
variable when the moderator is adjusted for. Therefore three models for each independent variable were
computed. First (model 1), sociodemographic factors were entered in the model followed by CHD-related
factors (model 2) and illness perceptions (model 3). This allowed us to examine whether including CHD
illness perceptions and attributions increases the share of explained variance of the model and whether the
main effects of CHD severity indicators would be reduced when illness perception and attributions are
accounted for. In the follow-up analysis the same procedure was followed using follow-up global health
status and QOL ratings as dependent variables and adjusting for baseline values of these variables.
Finally, to examine in more detail the paths through which the illness perceptions possibly mediate the
relationships between CHD severity and global health status and QOL, structural equation modelling (SEM)
was applied to perform a path analysis. The theoretical model appears in Figure 1. This proposed model
tested included three exogenous variable for CHD severity: CHD comorbidity, CHD risk factors and use of
nitrates. The preliminary analysis showed that fitting a latent factor measurement model for these CHD
severity indicators was unsuccessful due to low intercorrelations of the indicators. Therefore CHD severity
indicators were used as separate independent variables in the proposed model. The mediators were two latent
factors: ‘‘Attributions’’ which was composed of three observed variables (attribution to stress, risk factors
and internal factors), and ‘‘Illness Perceptions’’ which was composed of four observed variables (identity,
consequences, cure/control and time-line). The dependent variables were self-rated global health status and
global QOL in baseline and follow-up. Both direct effects of exogenous variables and indirect effects
mediated by illness perceptions and attributions were tested. Partial covariance matrix (adjusted for
sociodemographic factors, MI and revascularisation) was used to perform structural modeling.
The analyses were performed using the SAS (version 9.1) and SPSS (version 12.0.2) statistical packages.
AMOS 4.01 was used to perform SEM to perform confirmatory factor analyses and structural equation
modelling.
This is a NIVEL certified Post Print, more info at http://www.nivel.eu -5-

1 The Chow test is used to assess whether estimating parameters separately for sub samples significantly increases amount of the
explained variance compared to use of pooled sample. Based on the residual sum of the squares of the models computed for sub
samples (such as males and females) and for the pooled sample, the Chow test statistic tests the null hypothesis that the parameter
estimates of the models in the sub sample are the same (H0: b1=b2). controlling for the number of possible interaction terms. For the
null hypothesis, the Chow test-statistic (F = [(RSSR)SSR1)SSR2)/k]/[(SSR1 + SSR2)/ (n)2k)], where RSSR is the sum of squared
residuals from the pooled sample, SSR1 and SSR2 the corresponding sums in the two sub samples, and k the number of parameters
in the equation) has an F-distribution with k and (n)2k) degrees of freedom. Where the F-value was greater than the critical value for
F(k, n)2k) at 99% risk level, the significant gender interaction terms were examined individually.

RESULTS
Table 1 presents the characteristics of the sample. Approximately a third of respondents had suffered MI,
36% had undergone revascularisation and less than 20% used short-term nitrates on a weekly basis. Female