Psychosomatics
Journal Home Search Current Issue Past Issues Subscribe All APPI Journals Help Contact Us
 
Quicksearch
Advanced Search
Or Search All APPI Journals
This Article
* Abstract Freely available
* Full Text (PDF)
* Alert me when this article is cited
* Alert me if a correction is posted
* Citation Map
Services
* Email this article to a Colleague
* Similar articles in this journal
* Similar articles in PubMed
* Alert me to new issues of the journal
* Add to My Articles & Searches
* Download to citation manager
* reprints & permissions
Citing Articles
* Citing Articles via HighWire
* Citing Articles via Google Scholar
Google Scholar
* Articles by Höfer, S.
* Articles by Doering, S.
* Search for Related Content
PubMed
* PubMed Citation
* Articles by Höfer, S.
* Articles by Doering, S.
Related Collections
* Anxiety Disorders (General)
* Depression
* Syndromes Secondary to General Medical Disorders
* Symptoms/Dimensions
Psychosomatics 46:212-223, June 2005
© 2005 The Academy of Psychosomatic Medicine

Determinants of Health-Related Quality of Life in Coronary Artery Disease Patients: A Prospective Study Generating a Structural Equation Model

Stefan Höfer, Ph.D., W. Benzer, M.D., H. Alber, M.D., E. Ruttmann, M.D., M. Kopp, Ph.D., G. Schüssler, M.D., and S. Doering, M.D.

Received Oct. 7, 2003; revision received Feb. 17, 2004; accepted Aug. 12, 2004. From the Departments of Medical Psychology and Psychotherapy, Cardiology, Heart Surgery, and Psychiatry, Medical University Innsbruck, Innsbruck, Austria; the Department of Psychology, Royal College of Surgeons in Ireland, Dublin; the Department of Interventional Cardiology, Academic Hospital, Feldkirch, Austria; and the Departments of Prosthodontics, Psychosomatics and Psychotherapy, University of Muenster, Germany. Address correspondence and reprint requests to Dr. Höfer, Department of Medical Psychology and Psychotherapy, University of Innsbruck, Sonnenburgstr. 9, A-6020 Innsbruck, Austria; stefan.hoefer{at}uibk.ac.at (e-mail).


  ABSTRACT

 
TOP
 ABSTRACT
INTRODUCTION
 METHOD
 RESULTS
 DISCUSSION
 CONCLUSION
 REFERENCES
 
The authors used structural equation modeling to test a conceptual model of HRQL in coronary artery disease. The model, which included biomedical factors and individual and environmental characteristics, was tested in a multicenter group of 465 patients at three timepoints (baseline evaluation of chest pain and 1- and 3-month follow-ups). A satisfactory fit was obtained for the model over time. Depression and anxiety symptoms exerted the most significant influence on HRQL. HRQL and the mediating factors were found to be distinct phenomena. The authors concluded that mediating factors, especially depression and anxiety symptoms, should be taken into consideration in clinical routine if HRQL is regarded as a clinical outcome.

Key Words: Health related quality of life • cardiac disease • structural equation modeling • anxiety • depression • outcome

  INTRODUCTION

 
TOP
 ABSTRACT
 INTRODUCTION
METHOD
 RESULTS
 DISCUSSION
 CONCLUSION
 REFERENCES
 
Health-related quality of life (HRQL) is a widely accepted and frequently used outcome measure in clinical trials and health services research. The significance of the concept is documented by two journals exclusively focusing on HRQL.1,2 Despite its popularity in various fields of medicine, the concept of HRQL lacks theoretical background. The recent boom of HRQL research has been supported by the development of generic HRQL questionnaires—such as the Medical Outcomes Study 36-item Short-Form Health Survey (SF-36)3 and the Sickness Impact Profile4—and of disease-specific HRQL instruments, including the European Organisation for Research and Treatment of Cancer QOL Core Questionnaire 305 and the MacNew Heart Disease Questionnaire.6–8 In addition, interviewer-administered individualized instruments such as the Schedule for the Evaluation of Individual Quality of Life9 and computer-based versions of established HRQL instruments such as the Seattle Angina Questionnaire10 are available. However, during the last decade, research on HRQL focused primarily on assessment and has not contributed extensively to theory development or to adding empirical findings to an overall theoretical model. Thus, the history of the concept of HRQL tends to parallel research on the assessment of psychological well-being. In 1985 Abbey and Andrews11 stated that 15 years of psychometric effort had been put into finding effective ways to measure peoples’ sense of well-being, but the underlying question of how people come to feel as they do about their well-being was greatly neglected. They pointed out the need to link "well-being" to other established concepts in psychology, such as locus of control.

In 1995, Wilson and Cleary12 provided a conceptual model of HRQL that moved beyond observation of health status toward assessment of causal relationships among components of HRQL. Although by 2003 the Wilson and Cleary model had been cited more than 300 times,13 it was derived from theory and has not yet been fully tested empirically. Wilson and Cleary suggested specific causal relationships between health concepts encompassing biological, social, and psychological variables (Figure 1). A linear progression without dominant reciprocal effects or links between nonadjacent concepts was proposed. The five levels of the model are 1) biological-physiological variables, 2) symptom status, 3) physical functional status, 4) general health perception, and 5) quality of life. In addition, individual and environmental factors are included as unspecific mediating variables. Although this model outlines potential causal relations between the variables that play a major role in the origins of HRQL, little is known about the details of the interaction of these variables and their temporal stability. For example, Sullivan and colleagues14 tried to identify the proper place of anxiety and depression in this model. They concluded that anxiety and depression as moderating variables determine overall quality of life as strongly as perceived health. In addition, mental health status affects quality of life most potently in its shaping of one’s appraisal of his or her health status. However, their study subjects were limited to elderly Dutch people, and their findings need replication.



View larger version (12K):
[in this window]
[in a new window]
 

FIGURE 1.  Conceptual Model of Health-Related Quality of Lifea

aBased on the theoretical work of Wilson and Cleary.12


The assessment of outcome in cardiac disease still relies primarily on a model that targets biological parameters and mortality and fails to cover more recently developed biopsychosocial concepts. Although measurement of HRQL in cardiac disease has become more accepted during the last few years15 and can be considered a relevant endpoint,16 effort was put mainly into identifying the psychosocial factors that cause or influence the course of the disease,17 and the influence of these factors on HRQL as an outcome has not been examined.

To our knowledge, only one study has tried to identify variables that determine perceived health in patients with cardiac disease. Rosen and colleagues18 examined whether emotional distress, social support, and physical functioning, together with sociodemographic and clinical variables, could predict subjective global health in patients with left ventricular dysfunction. They found that sociodemographic and clinical variables had no or only indirect effects on quality of life. Social support and emotional distress were related to physical functioning. However, as the authors themselves stated, the study had some methodological limitations, and the findings need replication.

Testing a single predictor in a cross-sectional design may overestimate the influence of the tested predictor or may not detect its mediating effects. Therefore, in this study, we included several variables that were shown to have a significant influence on HRQL in previous investigations. To test simultaneously the underlying relationships between the large number of variables, we employed a structural equation modeling approach. This approach is recommended for the development of models in HRQL research,19 because it allows for the simultaneous evaluation of a set of measurement models and structural path coefficients. In such models, latent constructs such as physical functional status or global HRQL are defined by measured variables or by a combination of two or more measured variables. Structural path coefficients (regression coefficients) indicate the association between latent constructs or between single predictors and latent constructs. To ensure the validity and reliability (stability) of the model, we used a longitudinal approach covering three assessments during 3 months.

The main goal of this study was to apply the model of Wilson and Cleary12 a priori to patients with coronary artery disease in a prospective longitudinal design and to find out whether it is applicable to coronary artery disease patients and is stable over time.


  METHOD

 
TOP
 ABSTRACT
 INTRODUCTION
 METHOD
RESULTS
 DISCUSSION
 CONCLUSION
 REFERENCES
 
Patients
We recruited 465 patients in two hospitals in the western region of Austria—367 patients at the Department of Cardiology, Medical University Innsbruck, and 98 at the Department of Interventional Cardiology, Academic Hospital, Feldkirch. All patients were referred to the cardiology departments for angiographic evaluation of chest pain. Patients who had experienced an acute myocardial infarction (MI) during the last 6 months or had documented heart failure were not approached. Signed informed consent, knowledge of the German language, angiographically documented coronary artery disease, absence of other chronic diseases, and absence of severe mental disorder were inclusion criteria. Trained interviewers approached the patients on the day of admission before angiography, and complete data were collected. Of the initial 465 patients, 33 had no angiographically documented coronary artery disease, and their data were dropped from further analysis. The remaining 432 patients were included on an intention-to-treat basis. Treatment decisions were based on the number of diseased vessels, extent of left ventricular function, and clinical symptoms and were made by experienced cardiologists. Treatments included appropriate medication and, in addition, either percutaneous coronary intervention or coronary artery bypass graft surgery, if appropriate. Continuous medical treatment was prescribed as a stand-alone intervention or with a combination of beta-blockers, calcium antagonists, angiotensin-converting enzyme inhibitors, statins, aspirin, nicorandil, and nitrates. All patients were followed up at 1 month and 3 months after the initial treatment. The 1-month follow-up was completed by 305 patients (70.6%), and 277 (64.1%) returned for the 3-month follow-up. Of the 432 patients, four died for reasons related to cardiac disease, 27 withdrew their informed consent, 103 refused further participation, and 21 were not available for follow-up. Because it was not the primary aim of this analysis to assess change of HRQL over time, data for all available patients were included in the analyses at each timepoint.

Measures
For testing the model of Wilson and Cleary,12 single-domain and multiple-domain indicators of the five proposed latent variables were identified in the first step. Some existing HRQL measures include several concepts as indicators of latent variables, and others only one.12 The latent and measured variables8,20–26 analyzed in the present report are summarized in the following list and in Table 1:


View this table:
[in this window]
[in a new window]
 

TABLE 1. Latent Variables and Measured Variables in a Prospective Study of Health-Related Quality of Life (HRQL) in Coronary Artery Disease Patients



1) Latent variable: biological and physiological status; measured variables: number of coronary arteries that were diseased and number of risk factors (e.g., hyperlipidemia, diabetes mellitus, etc.). A vessel was considered to be significantly diseased if more than 50% lumen stenosis was demonstrated in one or more major branches of the coronary tree. Interviewers identified the risk factors by asking the patient about risk factors and consulting clinicians’ records of routine assessments.

2) Latent variable: symptom status; measured variable: Canadian Cardiovascular Society classification of angina pectoris.20

3) Latent variable: physical functional status; measured variable: SF-36 physical function scale score.21

4) Latent variable: general health perception; measured variable: SF-36 general health perception score.21

5) Latent variable: quality of life; measured variable: scores on the three scales (physical, social, and emotional HRQL) of the MacNew Heart Disease Quality of Life Questionnaire.8 In contrast to Wilson and Cleary, whose model included general quality of life, we assessed disease-specific global quality of life by using the MacNew Heart Disease Quality of Life Questionnaire.

In the second step, we attempted to identify indicators for the proposed individual and environmental variables from the model of Wilson and Cleary12 and to choose suitable instruments for the assessment of these variables. The assessment instruments are described in the following sections.

MacNew Heart Disease Quality of Life Questionnaire
The self-administered MacNew Heart Disease Quality of Life Questionnaire,6,27 designed to evaluate how daily activities and emotional and social functioning are affected by coronary artery disease, is based on an instrument originally developed with a focus-group approach and validated in English for patients with myocardial infarction.28,29 The MacNew questionnaire has been shown to be valid and reliable in English for patients with myocardial infarction,6,27 and in Spanish and Dutch, with reference norms available for patients after myocardial infarction, angina, and heart failure.30 In earlier work, the MacNew questionnaire was translated into German by using accepted linguistic validation forward- and back-translation techniques and has been validated for use with patients with angina pectoris.8

The MacNew questionnaire has a 2-week timeframe and contains 27 items, some of them falling into more than one domain. They are scored from 1 (poor) to 7 (high), with a physical limitations scale of 13 items, an emotional function scale of 14 items, and a social function scale of 13 items. The questionnaire includes seven questions about symptoms (two questions about angina/chest pain and one question each about shortness of breath, fatigue, dizziness, aching legs, and restlessness).6,27

SF-36
The SF-36 is a valid and widely used generic HRQL questionnaire that has been translated and adapted for use in more than 40 countries.31 The instrument includes 36 items that are combined in eight scales: physical functioning, role—physical, body pain, general health, vitality, social functioning, role—emotional, and mental health. These scales can be combined into two component scores, and there is one item about present health status (health transition). The German version of the SF-36 has been shown to be valid and reliable.21

Social Support
There is evidence that social support plays an important role in the onset of disease and in the recovery process, especially after myocardial infarction.17,32–34

In a healthy population, social support also plays an important role in handling stressful events with respect to cardiac health. Moreover, it is not social support itself but the way social support is perceived by the individual that influences whether social support is helpful.35 There is also evidence that social support influences quality of life in coronary artery disease patients.36

Most studies investigating structural measures of social support focused on the extent and stability of the social network, but little research has discriminated between different types of functional support (e.g., providing information, tangible support, emotional support) and satisfaction with social support.

We assessed social support by means of the German short form of the Social Support Questionnaire.24,25 It includes 22 items and covers a broad range of domains of social support, including emotional and practical support, social integration, availability of a close confidant, and satisfaction with social support. The internal consistency of the scales ranges from 0.70 to 0.92 (Cronbach alpha), with the short form showing good correlation with the long form (r=0.66–0.98).24,25

Locus of Control
Another variable that has been shown to have an effect on global quality of life and HRQL is locus of control.11 This variable focuses on expectancy beliefs for future events and can be linked to health. It refers to individuals’ beliefs about who or what determines outcomes in their lives. Several researchers have hypothesized that perceptions of internal control are associated with well-being and therefore with quality of life.11 It is assumed that locus of control has an effect on the outcome of chronic disease.37 At least one study that investigated the influence of locus of control on the recovery process in coronary artery disease patients found that locus of control was a predictor of recovery after myocardial infarction.38

In this study, locus of control was regarded as a personal resource and an individual characteristic (personality trait) that influences patients’ perceptions of HRQL. For the assessment of locus of control, the German version of the Competence and Control Orientations Questionnaire26 was used. The 32 items of the questionnaire are combined in four scales: self-concept (generalized expectations of the availability of possible activity in life events), internal locus of control (general belief that one is in control of one’s own life), powerful others (generalized expectations that one’s life is dependent on powerful others), and chance control orientation (general belief that one’s life is mainly influenced by chance, luck, or misfortune). In contrast to other measures of locus of control, the Competence and Control Orientations Questionnaire assumes multidimensionality of the locus-of-control personality trait. Internal consistency of the scales ranged from 0.70 to 0.89 (Cronbach alpha), with good test-retest reliability (r=0.63–0.92).26

Anxiety and Depression
Anxiety and especially depression are well-known risk factors for onset of and relapse in coronary artery disease17 and are predictors of the recovery process after myocardial infarction.32 In addition, they are regarded as influencing factors for HRQL in cardiac disease patients.39,40

Although Wilson and Cleary12 argued that depression or any other psychological factor may be regarded as a biological and physiological variable, as a symptom status variable, or as a functioning status variable, we assessed anxiety and depression as individual characteristics that may shape the appraisal of health status.14

Anxiety and depression were measured by means of the German version of the Hospital Anxiety and Depression Scale.22 Because the Hospital Anxiety and Depression Scale is primarily intended for the assessment of pathological anxiety, anxiety as a trait was in addition assessed with the German version of the State-Trait Anxiety Inventory.23

Data Analysis
Structural equation modeling was used to test the theoretical model against the observed dataset. Structural equation modeling is a combination of factor analysis and path analysis and is described in detail elsewhere.41–44 It is a confirmatory rather than an exploratory technique, because it compares a hypothesized model’s covariance matrix with that of the observed data. Typically, this approach allows a more "causal" explanation of findings. There are several steps that need to be considered when applying structural equation modeling: 1) developing a model based on theory; 2) identification of unique values that can be used for the parameters to be estimated in the theoretical model; 3) application of various estimation techniques, for example, maximum likelihood; and 4) testing the fit of the model against the data. According to the results, the researcher might 5) modify the measurement model based on theoretical justifications; revise the model by adding, deleting, or modifying relationships between latent variables; or use measures indicating lack of fit for specific parts of the model when theoretically justified.45 Compared with standard regression methods, structural equation modeling is a more theory-driven approach, and the resulting prediction equations are a more accurate representation of the true causes of variation in the dependent variable.

Structural equation modeling allows for simultaneous evaluation of a set of measurement models and path coefficients. Latent constructs such as HRQL are assessed by two or more measured variables. Structural path coefficients reflect associations between latent constructs or between single-indicator predictors and latent constructs. Structural equation modeling encompasses two major components: 1) measurement models (e.g., confirmatory factor analysis) and 2) structural path components (e.g., regression analysis). In our analysis we used both measurement models and structural path components to build a full latent variable model,42 or hybrid model.41

Before the full latent variable model was tested, each measurement model (e.g., social support, locus of control) included in the full model was tested separately to ensure its fit, by using the two-step approach recommended by Anderson.46 This process involved an evaluation of the hypothesis that the indicated measured items or scales reflect the latent constructs. Models for each construct were defined by permitting each of the relevant test items or scale scores to load on a single factor representing the latent construct that it was hypothesized to measure.

Goodness of fit indices were used as an indicator of model fit. Chi-square tests were used as an index of the significance of the discrepancy between the original (sample) correlation matrix and the (population) correlation matrix estimated from the model. Because the significance of chi-square tests is dependent on the number of subjects, the comparative fit index (CFI) and the root mean square error approximation (RMSEA) were further considered. CFI values are derived from the comparison of the hypothesized model with the independence model. RMSEA values help to answer the question of how well the model—with unknown but optimally chosen parameter values—would fit the population covariance matrix if it were available.42 The lower the discrepancy measured by the RMSEA the better, with an RMSEA of 0.0 indicating a perfect fit. Acceptable values are CFI >.90 and RMSEA <.08. For the comparison of models, we used the chi-square statistic.

Analyses were conducted by using Amos 4.043 and SPSS 11.0.47


  RESULTS

 
TOP
 ABSTRACT
 INTRODUCTION
 METHOD
 RESULTS
DISCUSSION
 CONCLUSION
 REFERENCES
 
Patients’ Characteristics
Testing and development of the model was conducted by using baseline data. The characteristics of the 432 patients available at baseline before intervention are presented in Table 2. At 1-month and 3-month follow-up, 70.6% (N=305) and 64.1% (N=277) of the original patients, respectively, were available. The number of dropouts differed significantly between groups receiving different treatment methods ({chi}2=24.18, df=2, p<0.001). Sixty-three (35.2%) patients receiving continuous medical treatment and 66 (50.8%) patients receiving coronary artery bypass graft dropped out, but only 26 (21.1%) patients in the percutaneous coronary intervention group did not complete both follow-ups. However, the patients who remained in the study did not differ significantly in any clinical or sociodemographic variable from the dropouts.


View this table:
[in this window]
[in a new window]
 

TABLE 2. Characteristics of Coronary Artery Disease Patients (N=432) in a Prospective Study of Health-Related Quality of Life



HRQL Model
Measurement errors for the measured variables were allowed to intercorrelate where appropriate (e.g., social support). Because the model of Wilson and Cleary12 did not specify in detail the external variables that influence different health statuses, we allowed all individual and environmental variables to load on all health statuses in our first structural equation model (see Table 3). This model implies that all chosen individual and environmental characteristics equally influence all levels of the variables of the directional model. This model did not show satisfying statistical parameters ({chi}2=2391.42, df=168, p<0.0001; CFI=0.75, normed fit index [NFI]=0.70, RMSEA= 0.19), indicating that the chosen individual and environmental variables did not equally influence the identified levels of health status. This result also made it clear that, in addition to identifying mediating variables, the part of the model modified by the individual and environmental variables must be specified.


View this table:
[in this window]
[in a new window]
 

TABLE 3. Goodness of Fit of Structural Equation Models of Variables in a Prospective Study of Health-Related Quality of Life (HRQL) in Coronary Artery Disease Patients



A second model was developed by maintaining significant paths and using modification indices (see Figure 2). This model included only the significant paths and underlying relationships of the mediating variables. The chi-square test indicated that this model was of less-than-perfect fit ({chi}2=513.28, df=188, p<0.001). However, the chi-square difference test revealed that the final model fit the data better than its corresponding null model ({chi}2=3574.37, df=42, p<0.001). The null hypothesis associated with this model assumed that there is no covariation among the variables in the model. Null models can be used as standards against which to evaluate the adequacy of substantive empirical models. In addition, the goodness of fit indices showed acceptable values (NFI=0.88, CFI=0.92, RMSEA=0.06) (Table 3). Parameter estimates derived from the final model are presented in Figure 2. These coefficients represent complete standardized maximum likelihood estimates of factor loadings and structural path coefficients. Not shown in Figure 2 are the correlations between measurement errors to ensure better visibility of the model. The final model explains 49% of variance occurring in global HRQL at baseline.



View larger version (35K):
[in this window]
[in a new window]
 

FIGURE 2.  Significant Pathways in the Final Health-Related Quality of Life Modela

aStandardized coefficients indicating relationships between variables are shown close to the relevant path. The first coefficient indicates the baseline relationship, the second indicates the relationship at 1 month follow-up, and the third indicates the relationship at 3-month follow-up.


Structural path coefficients representing predictive relationships between the biological and physiological variables (number of risk factors and number of diseased vessels) and symptom status indicated a weak but expected significant directional relationship. The more vessels that are involved (ß=0.15) and the more risk factors present (ß=0.08), the worse the symptom status, that is, patients reported more symptoms or felt more restricted by their symptoms if more vessels were diseased. Symptom status in turn negatively influenced physical functioning status measured by the SF-36 physical functioning scale. The worse the symptom status reported, the less patients were able to perform physical activities (ß=–0.09). Moreover, limitation in physical functioning led to a worse general health perception (ß=0.25). General health perception had the predicted significant directional influence on global HRQL (ß=0.24).

As shown in Figure 2, several significant associations were observed between individual and environmental characteristics on the one hand and global HRQL on the other hand. The highest effect on global HRQL was exerted by physical functioning (ß=0.36) and anxiety symptoms (ß=–0.36). The more anxious patients were, the poorer the perception of global HRQL. Limitation in physical functioning led to a reduced global HRQL. The other mediating variables did not show a direct influence on global HRQL. However, all psychopathological variables (depression, anxiety symptpoms, and trait anxiety) showed a high relationship with emotional HRQL as part of the global HRQL (ß=–0.12, ß=–0.23, and ß=–0.06, respectively). In addition, depression had a considerable effect on physical functioning status (ß=–0.32) and general health perception (ß=–0.32). The presence of anxiety symptoms showed a weak but significant positive relationship with symptom status (ß=0.10), indicating that the higher the level of anxiety, the more severe the symptoms reported. Trait anxiety showed a negative relation to general health perception, indicating that anxiety as a trait negatively influenced the perception of general health (ß=–0.19).

Among the demographic variables, only age and gender were included in the second model; both showed a modest relationship with physical functioning status (ß=–0.14 and ß=–0.12, respectively). Female patients and older patients were more likely to report impairment of their physical abilities.

With regard to locus of control, only powerful others and chance control (ß=0.20 and ß=–0.14, respectively) showed a moderate relation to general health perception. People who had the belief that powerful others influence outcome and those who were less likely to believe in chance reported a higher general health perception. In addition, a weak relationship between a low self-concept and a low emotional HRQL was detected (ß=0.06). However, the other locus-of-control variables showed no significant influence on any of the postulated variables in the model.

Within the concept of social support, only satisfaction with support showed a weak relationship with general health perception (ß=0.09).

Additional relationships were discovered between individual and environmental variables. A high level of depression was linked to low self-concept (ß=–0.10) and to a low level of feeling socially integrated (ß=–0.19). Anxiety symptoms predicted a significantly worse self-concept (ß=–0.11).

In a third step of analyses, the same final model was applied to the data from the follow-up investigations 1 month and 3 months after the intervention. One and 3 months after the intervention, the indices showed an acceptable fit (NFI=0.82 and NFI=0.86, respectively; CFI=0.91 and CFI=0.91, respectively; RMSEA=0.08 and RMSEA=0.08, respectively), explaining 62% and 66%, respectively, of the variance of global HRQL. Standardized coefficients for the follow-up data are shown in Figure 2. In general, the standardized coefficients remained stable.

The influence of anxiety symptoms on symptom status increased over time (from ß=0.32 at 1 month to ß=0.44 at 3 months). One month after the intervention, depression had a high level of influence on the perception of physical functioning (ß=0.40), but this influence decreased by 3 months (ß=–0.28).

The direct influence of physical functioning on global HRQL increased at 1-month follow-up (ß=0.48) and then decreased at 3 months (ß=0.20). Some path coefficients that were significant at baseline failed to reach the significance level at the 1-month and 3-month follow-ups, including powerful others, chance control, and number of diseased vessels.

In addition to examining the direct effects represented by the paths within the model, we also investigated indirect effects of the variables on global HRQL. At baseline, female gender (ß=–0.05), older age (ß=–0.06), depression (ß=–0.21), and limitation in physical functioning (ß=0.06) influenced global HRQL negatively. One month after the intervention, the indirect influence of depressive symptoms weakened (ß=–0.05) and age lost its indirect effect on global HRQL (ß=0.01); the influence of female gender (ß=–0.14) and symptom status (ß=–0.07) increased.

However, after 3 months, depression again showed the highest indirect influence on global HRQL (ß=–0.22). Symptom status (ß=–0.15), physical functioning (ß=0.10), gender (ß=–0.10), anxiety symptoms (ß=–0.07), and social support (ß=0.05) also showed indirect but weak effects on global HRQL at 3 months.


  DISCUSSION

 
TOP
 ABSTRACT
 INTRODUCTION
 METHOD
 RESULTS
 DISCUSSION
CONCLUSION
 REFERENCES
 
The model of Wilson and Cleary12 linking individual and environmental variables to global HRQL was examined in this study. To our knowledge, our study is the first to outline the pattern of associations between individual and environmental characteristics and global HRQL in patients with coronary artery disease by using structural equation modeling in a longitudinal design. The final model presented here linked clinical variables, such as the number of diseased vessels and the number of risk factors, to global HRQL through the mediating effects of the experience of actual symptoms (i.e., symptom status), physical functioning, and general health perception. The use of structural equation modeling permitted simultaneous evaluation of the effects of individual and environmental characteristics on the latent variables within the model. In addition, indirect effects could be assessed, which would have been impossible with standard regression methods. The overall model explained at least 49% of the variance in subjective global HRQL, a result that supports the use of structural equation modeling methods in the investigation of the perception of HRQL.

Although Wilson and Cleary proposed a linear model without dominant reciprocal effects or links between nonadjacent concepts, we allowed a direct effect of the physical functioning status on the perception of global HRQL, guided by modification indices. By revealing the different patterns of variables influencing the model, our results confirmed the distinction of health status and HRQL that was made by Smith et al.19 However, in our model, health status played a major influential role on global HRQL.

Because the final model was tested at three separate time points and showed an acceptable fit each time, its validity was clearly demonstrated. Although minor changes of the regression weights occurred, the value of the regression weights for the individual and environmental variables that influenced HRQL remained highly stable over time.

Among the individual variables, anxiety symptoms and depression showed the greatest effect on the linked variables in the model. Although depression significantly influenced the emotional aspect of HRQL, it had no significant direct effect on global HRQL. However, depression had a major and constant effect on the perception of physical functioning and general health, exerting a major indirect effect on global HRQL at both baseline and 3 months. This depression effect had almost disappeared at 1 month after initial treatment but reoccurred after 3 months. This result suggests that subjective evaluation of global HRQL shortly after treatment is influenced more by the perception of physical functioning, but later on, the indirect effects of variables such as depression regain influence on the perception of HRQL. Thus, the results suggest that assessment of HRQL is distinct from that of depression but that depression represents the most important indirect influence on the course of HRQL in coronary artery disease patients.

Anxiety symptoms were the second major influencing individual parameter and the only individual and environmental variable that showed a high and constant direct effect on the perception of global HRQL. Unlike the relationship of depression and HRQL, the relationship of anxiety symptoms and HRQL did not show reverse developments during our investigation. Thus, theoretically it cannot be ruled out that there may exist some overlap between the concepts and the measurement of anxiety symptoms and HRQL.

Our findings that anxiety symptoms and depression have major effects on global HRQL in cardiac disease patients extend the findings of Sullivan et al.,14 who suggested that anxiety and depression mediate the relation between general health perceptions and overall quality of life in a general Dutch population. In contrast to their conclusion, our findings suggest that anxiety and depression do not mediate between general health perception and global HRQL but influence the perception of physical functioning, general health, and/or global HRQL directly.

Social support did not have much influence within the model, except on general health perception. The importance of this effect increased with time, which suggests that social support is less important in coping with acute disease but becomes more important the longer the state of illness continues.

Age and gender as individual characteristics showed a constant effect on physical functioning but no direct effect on global HRQL. Female patients and older patients appeared to perceive a lower level of ability to perform physical tasks.

Self-esteem, internal control, self-efficacy, personal autonomy, and perceived control are often referred to as crucial for the perception of control.48 However, we found only limited influence of control beliefs within the HRQL model. Again, general health perception was influenced directly, surprisingly not by internal locus of control, but by the belief in powerful others (positive effect) and chance control (negative effect). Our interpretation of this result is that patients admitted to a hospital profit from an ability to surrender responsibility and control to the doctors, as the situation encourages them to believe that these powerful others have a role in the outcome. On the other hand, distrustful patients who believe in chance rather than the doctors’ competence are rather prone to perceive their general health as worse. This interpretation follows the suggestion of Cohen and Lazarus,49 who argued that in hospitalized patients, particularly in a perioperative context, a vigilant coping style with its intention to maintain internal control is inferior to a style of surrendering responsibility to the doctors in this situation of forced dependency and passivity. Our findings substantiate and extend the results of Abbey and Andrews,11 who suggested that internal and external locus of control have a significant but minor influence on the perception of global life quality.

A potential limitation of this study was the relatively low response rate at the follow-up investigations. These low rates might have produced a bias toward higher satisfaction, better physical functioning status, better health perception, and better HRQL. However, if this bias was present, the regression coefficients that we observed would represent minimum values that likely would increase if a larger number of subjects had been available at 1 and 3 months after the intervention. However, as our primary goal was not the assessment of change of HRQL after treatment, we assume that our observed regression weights represent reliable values for actual relationships 1 and 3 months after initial treatment. An ongoing 1-year follow-up will provide further data on the long-term validity of the model.


  CONCLUSION

 
TOP
 ABSTRACT
 INTRODUCTION
 METHOD
 RESULTS
 DISCUSSION
 CONCLUSION
REFERENCES
 
Our results provide empirical evidence for Wilson and Cleary’s theoretically derived model of HRQL. The five steps between symptom status and HRQL are significantly linked in the hypothesized direction, although there is a major direct influence of physical functioning on global HRQL. Moreover, individual and environmental factors show specific influences on the five central variables. Among the individual variables, depression and anxiety symptoms play the most important roles as mediator variables in the process toward HRQL. Social support and locus of control play minor, but still significant roles as mediator variables. Another interesting issue is the decrease in the indirect influence of depression immediately after the initiation of treatment. We conclude that symptoms and physical status in the acute phase of treatment (especially in patients receiving percutaneous coronary intervention and coronary artery bypass graft) have a greater effect on HRQL than individual factors and that, in the course of time, the mediating factors become more important.

The findings from the HRQL model presented here suggest that HRQL and the mediating factors can be regarded as distinct phenomena. The mediating factors influence HRQL, but they are not part of HRQL themselves. Thus, it is justified to regard and assess HRQL as a distinct psychological entity. With regard to HRQL in clinical routine, we conclude that mediating factors, especially depression and anxiety symptoms, should be taken into consideration and adequately treated if present.


  ACKNOWLEDGMENTS

 
This work was funded partly by grants to Dr. Höfer from Medical University Innsbruck (GZ: 14031/13–02) and the Department of Science, Government of Vorarlberg, Austria (IIb-11.01/0083).


  REFERENCES

 
TOP
 ABSTRACT
 INTRODUCTION
 METHOD
 RESULTS
 DISCUSSION
 CONCLUSION
 REFERENCES
 

  1. Bradley C, Tamburini M: Not-only-a-title. Health Qual Life Outcomes 2003; 1:1–10[CrossRef][Medline]
  2. Health-related quality of life research (editorial). Qual Life Res 1992; 1:3[CrossRef][Medline]
  3. Tarlov AR, Ware JE Jr, Greenfield S, Nelson EC, Perrin E, Zubkoff M: The Medical Outcomes Study: an application of methods for monitoring the results of medical care. JAMA 1989; 262:925–930[Abstract]
  4. Bergner M, Bobbitt RA, Carter WB, Gilson BS: The Sickness Impact Profile: development and final revision of a health status measure. Med Care 1981; 19:787–805[Medline]
  5. Fayers P, Bottomley A: Quality of life research within the EORTC—the EORTC QLQ-C30. European Organisation for Research and Treatment of Cancer. Eur J Cancer 2002; 38(suppl 4):S125-S133
  6. Lim LL, Valenti LA, Knapp JC, Dobson AJ, Plotnikoff R, Higginbotham N, Heller RF: A self-administered quality-of-life questionnaire after acute myocardial infarction. J Clin Epidemiol 1993; 46:1249–1256[CrossRef][Medline]
  7. Höfer S, Benzer W, Brandt D, Laimer H, Schmidt P, Bernardo A, Oldridge NB: MacNew Heart Disease Lebensqualitätsfragebogen nach Herzinfarkt: Die deutsche Version. Zeitschrift für Klinische Psychologie und Psychotherapie, Oktober 2004; 33:270–280[CrossRef]
  8. Höfer S, Benzer W, Schüssler G, von Steinbüchel N, Oldridge NB: Health-related quality of life in patients with coronary artery disease treated for angina: validity and reliability of German translations of two specific questionnaires. Qual Life Res 2003; 12:199–212[CrossRef][Medline]
  9. Hickey A, O’Boyle CA, McGee H, Joyce CRB: The Schedule for the Evaluation of Individual Quality of Life, in Individual Quality of Life: Approaches to Conceptualisation and Assessment. Edited by Joyce CRB, O’Boyle CA, McGee H. Amsterdam, Harwood Academic Press Publishers, 1999, pp 119–133
  10. Bliven BD, Kaufman SE, Spertus JA: Electronic collection of health-related quality of life data: validity, time benefits, and patient preference. Qual Life Res 2001; 10:15–22[CrossRef][Medline]
  11. Abbey A, Andrews FM: Modeling the psychological determinants of life quality. Soc Indicators Res 1985; 16:1–34
  12. Wilson IB, Cleary PD: Linking clinical variables with health-related quality of life: a conceptual model of patient outcomes. JAMA 1995; 273:59–65[Abstract]
  13. Web of Science Institute for Scientific Information, 2003 (www.isiknowledge.com)
  14. Sullivan MD, Kempen GI, Van Sonderen E, Ormel J: Models of health-related quality of life in a population of community-dwelling Dutch elderly. Qual Life Res 2000; 9:801–810[CrossRef][Medline]
  15. Swenson JR, Clinch JJ: Assessment of quality of life in patients with cardiac disease: the role of psychosomatic medicine. J Psychosomatic Res 2000; 48:405–415[CrossRef][Medline]
  16. Quality of life and clinical trials. Lancet 1995; 346(8966):1–2[Medline]
  17. Rozanski A, Blumenthal JA, Kaplan J: Impact of psychological factors on the pathogenesis of cardiovascular disease and implications for therapy. Circulation 1999; 99:2192–2217[Abstract/Free Full Text]
  18. Rosen RC, Contrada RJ, Gorkin L, Kostis JB: Determinants of perceived health in patients with left ventricular dysfunction: a structural modeling analysis. Psychosom Med 1997; 59:193–200[Abstract/Free Full Text]
  19. Smith KW, Avis NE, Assmann SF: Distinguishing between quality of life and health status in quality of life research: a meta-analysis. Qual Life Res 1999; 8:447–459[CrossRef][Medline]
  20. Cox JL, Naylor CD, Johnstone DE: Limitations of Canadian Cardiovascular Society classification of angina pectoris. Am J Cardiol 1994; 74:276–277[CrossRef][Medline]
  21. Bullinger M, Kirchberger I: SF-36 Fragebogen zum Gesundheitszustand. Göttingen, Germany, Hogrefe Verlag, 1998
  22. Herrmann C, Buss U: Vorstellung und Validierung einer deutschen Version der "Hospital Anxiety and Depression Scale" (HAD-Skala): ein Fragebogen zur Erfassung des psychischen Befindes bei Patienten mit körperlichen Beschwerden. Diagnostica 1994; 40:143–154
  23. Laux L, Glanzmann P, Schaffner P, Spielberger CD: Das State-Trait-Angstinventar. Weinheim, Germany, Beltz Testgesellscahft, 1981
  24. Franke GH: Testtheoretische Überprüfung des Fragebogen zur Sozialen Unterstützung. Zeitschrift für Medizinische Psychologie 1994; 3:168–177
  25. Sommer G, Fydrich T: Entwicklung und Überprüfung eines Frageboens zur sozialen Unterstützung. Diagnostica 1991; 37:160–178
  26. Krampen G: Fragebogen zu Kompetenz- und Kontrollüberzeugungen. Göttingen, Germany, Hogrefe Verlag für Psychologie, 1991
  27. Valenti L, Lim L, Heller RF, Knapp J: An improved questionnaire for assessing quality of life after acute myocardial infarction. Qual Life Res 1996; 5:151–161[CrossRef][Medline]
  28. Oldridge N, Guyatt G, Jones N, Crowe J, Singer J, Feeny D, McKelvie R, Runions J, Streiner D, Torrance G: Effects on quality of life with comprehensive rehabilitation after acute myocardial infarction. Am J Cardiol 1991; 67:1084–1089[CrossRef][Medline]
  29. Hillers TK, Guyatt GH, Oldridge N, Crowe J, Willan A, Griffith L, Feeny D: Quality of life after myocardial infarction. J Clin Epidemiol 1994; 47:1287–1296[CrossRef][Medline]
  30. Höfer S, Lim LL, Guyatt GH, Oldridge NB: The MacNew Heart Disease health-related quality of life instrument: a summary. Health Qual Life Outcomes 2004; 2:3[CrossRef][Medline]
  31. Aaronson NK, Acquadro C, Alonso J, Apolone G, Bucquet D, Bullinger M, Bungay K, Fukuhara S, Gandek B, Keller S, et al: International Quality of Life Assessment (IQOLA) Project. Qual Life Res 1992; 1:349–351[CrossRef][Medline]
  32. Jenkins CD, Stanton BA, Jono RT: Quantifying and predicting recovery after heart surgery. Psychosom Med 1994; 56:203–212[Abstract/Free Full Text]
  33. Ruberman W, Weinblatt E, Goldberg JD, Chaudhary BS: Psychosocial influences on mortality after myocardial infarction. N Engl J Med 1984; 311:552–559[Abstract]
  34. Case RB, Moss AJ, Case N, McDermott M, Eberly S: Living alone after myocardial infarction: impact on prognosis. JAMA 1992; 267:515–519[Abstract]
  35. Uchino BN, Holt-Lunstad J, Uno D, Flinders JB: Heterogeneity in the social networks of young and older adults: prediction of mental health and cardiovascular reactivity during acute stress. J Behav Med 2001; 24:361–382[CrossRef][Medline]
  36. Bosworth HB, Siegler IC, Olsen MK, Brummett BH, Barefoot JC, Williams RB, Clapp-Channing NE, Mark DB: Social support and quality of life in patients with coronary artery disease. Qual Life Res 2000; 9:829–839[CrossRef][Medline]
  37. Shaw C: A framework for the study of coping, illness behaviour and outcomes. J Adv Nurs 1999; 29:1246–1255[CrossRef][Medline]
  38. Rankin SH: Women recovering from acute myocardial infarction: psychosocial and physical functioning outcomes for 12 months after acute myocardial infarction. Heart Lung 2002; 31:399–410[CrossRef][Medline]
  39. Lane D, Carroll D, Ring C, Beevers DG, Lip GY: Effects of depression and anxiety on mortality and quality-of-life 4 months after myocardial infarction. J Psychosom Res 2000; 49:229–238[CrossRef][Medline]
  40. Oldridge N, Gottlieb M, Guyatt G, Jones N, Streiner D, Feeny D: Predictors of health-related quality of life with cardiac rehabilitation after acute myocardial infarction. J Cardiopulm Rehabil 1998; 18:95–103[CrossRef][Medline]
  41. Kline R: Principles and Practice of Structural Equation Modeling. New York, Guilford, 1998
  42. Byrne BM: Structural Equation Modeling With AMOS: Basic Concepts, Applications, and Programming. Mahwah, NJ, Lawrence Erlbaum Associates, 2001
  43. Arbuckle JL, Wothke W: Amos 4.0 User’s Guide. Chicago, Small Waters Corporation, 1999
  44. Brown M, Cudeck R: Alternative ways of assessing model fit, in Testing Structural Equation Models. Edited by Bollen K, Long J. Newbury Park, Calif, Sage Publications, 1993, pp 136–162
  45. Bollen K, Long J: Testing Structural Equation Models. Newbury Park, Calif, Sage Publications, 1993
  46. Anderson JCGDW: Structural equation modeling in practice: a review and recommended two-step approach. Psychol Bull 1988; 103:411–423[CrossRef]
  47. SPSS 11.0. Chicago, SPSS Inc, 2001
  48. Barry M: Well-being and life satisfaction as components of quality of life in mental disorders, in Quality of Life in Mental Disorders. Edited by Katschnig H, Freeman H, Sartorius N. Chichester, England, John Wiley & Sons, 1997, pp 31–42
  49. Cohen F, Lazarus RS: Active coping processes, coping dispositions, and recovery from surgery. Psychosom Med 1973; 35:375–389[Abstract/Free Full Text]



This article has been cited by other articles:


Home page
 Psychosom. Med.Home page
J. L. Levenson, D. K. McClish, B. A. Dahman, V. E. Bovbjerg, V. de A. Citero, L. T. Penberthy, I. P. Aisiku, J. D. Roberts, S. D. Roseff, and W. R. Smith
Depression and Anxiety in Adults With Sickle Cell Disease: The PiSCES Project
Psychosom Med, February 1, 2008; 70(2): 192 - 196.
[Abstract] [Full Text] [PDF]

This Article
* Abstract Freely available
* Full Text (PDF)
* Alert me when this article is cited
* Alert me if a correction is posted
* Citation Map
Services
*