ORIGINAL ARTICLES
Psychological Factors, Immunologic .Activation, and Disease Activity in Rheumatoid Arthritis Jerry C. Parker, Karen L. Smarr, Edgar 0. Angelone, Philip K. Mothersead, Byung S. Lee, Sara E. Walker, Alan 1. Bridges, and C. William Calclwell The purpose of this study was to use structural equation modeling techniques to examine potential interrelationships ctmong psychological factors, immunologic activation, and disease activity in rheumatoid arthritis (RA).The subjects were 80 male patients with a diagnosis of classic or definite RA. Measures included the Beck Depression Inventory, Ihe Arthritis Helplessness Index, and the Arthritis Impact Measurement Scales (AIMS) pain score. Joint counts and immunophenotypic analyses of periphrxal blood lymphocytes also were collected. Path cinalysis showed that percentage of HLA-DR cells in the peripheral Mood and helplessness were related to join count. In addition, joint count had an cffect upon depression. Depression had an effect upon pain. but there wus no reciprocal effect of pain upon depression. This study describes a preliminary path inodt:l of interrelationships among psychological fhctors, immunologic activation, and disease activ1 ty in RA. +
The interrelationships among psychological factors, Immunologic activation, and disease activity in rheumatoid arthritis (RA) constitute an area of conceptual
Jerrv C. Parker, PhI), Karen L. Smarr, BA, Edgar 0. Angelone, MA, Philip K. Mothersead. MA, Byung S. Lee, MA, Sara E. Walker, MD. Alan J. Bridges. !AD. and C. William Caldwell, MD, PhD, are with Harry S. Truman Memorial Veterans Hospital and University of Missouri-Columbia School of Medicine, Columbia, Missouri. Address correspondence to Jerry C. Parker, PhD, Psychology Service (116B),Harry :S. h u m a n Memorial Veterans' Hospital, 800 Hospital Drive, Colunibia, MO 65201. Submitted September 30, 1991; accepted March 31, 1992. 4 1992 by the Arthritis Hoalth Professions Association
196
and practical significance. Conceptually, there are important questions as to how RA disease activity may have an impact upon psychological states such as depression. Similarly, there are questions as to how R A disease activity relates to pain, because discrepancies between disease activity and selt-reported pain have been noted [1-3]. From a clinical perspective, an understanding of how symptoms such as depression, helplessness, and pain are manifested in RA is extremely important, but the methodological problems in this area of research are formidable. Several studies h a e found relationships between psychological variables and RA functional status [3-51, but studies using objective measures of disease activity are far less common. Even so, correlations between disease activity and psychological variables offer no information regarding causality. Psychological distress may exacerbate disease activity; but conversely, disease activity may exacerbate psychological distress. Reciprocal relationships are possible. The question of how psychological state and disease activity are related in RA has not been adequately answered by the correlational designs used to date. Eventually, data from prospective. randomized psychological intervention studies will be available, but an intermediate approach to exploring relationships between psychological state and RA disease activity involves structural equation modeling [6, 71. Structural equation modeling is a stiltistical technique that permits a researcher to postu1,ite an explanatory model and then to see if available data suppoirts the model. Structural equation modeling is a correlational technique, but involves a confirmatory (vs exploratory) statistical approach. Structural 0893-7'124192/$5 00
Psychological Factors 197
4rthritis Care and Research
equation modeling permits an analysis of how variables operate in combination. Therefore, structural equation modeling can play a particularly important role in the early stages of scientific inquiry. Hypothetical relationships can be examined in a preliminary way so thz t prospective, randomized studies can 3e designed more effectively. The purpose of this study was to use structural equation modeling techniques to examine potential interrelationships among psychological state, immunologic activ,ition, and disease activity in RA. Findings from prexious studies [3,8-121 contributed to the developnwnt of the proposed path model.
MATERIALS AND METHODS Subjects The subjects were 80 male patients from a midwestern Department ot Veterans Affairs hospital with a diagnosis of classic or definite RA. The diagnoses were made by a collaborating rheumatologist using the diagnostic criteria of the American Rheumatism Association [ 131 Exclusion criteria were as follows: (a) other uncontIolled medical disorders, (b) organic brain syndrome, [c) major psychiatric disturbance, (d) major communicative disorder, (e) history of severe noncompliance, (f) less than 7 years of education, and (g) illiteracy. The distribution of subjects by Steinbrocker Functional Class was as follows: Class I = 13%, Class I1 = 72%, and Class I11 = 15% [14]. The mean age for the sample was 61.6 (SD = 6.5) years. The mearl educational level was 11.0 (SD = 2.7) years. The mean Hollingshead index of socioeconomic status was 47.4. The mean disease duration for the sample was 13.4 years.
Measures Depression. The Beck Depression Inventory (BDI) was used to measure depression. The BDI is a 21item questionnaire that yields a single score [15].The BDI has good internal consistency, reliability, and validity [16]. Pain. The Pain Score from the Arthritis Impact Measurement Scales (AIMS) was used to measure pain [17]. The AIMS has been shown to be a reliable and valid instrument for measuring health status in RA patients [18 191. Helplessness. The Arthritis Helplessness Index (AHI) is a 15-item, self-report questionnaire that assesses patients’ perceptions of their ability to cope with
arthritis-related symptoms [lo].The AH1 yields a sin-gle score and has been shown to be a wliable and valid instrument [lo]. Disease Severity. Joint counts were used to measure disease activity. Joint counts were pix formed by an experienced rheumatology nurse clinician using the criteria of the American Rheumatism Association [ZO]. The procedure involved careful exarriination of each joint and a standardized quantification of painitenderness and swelling. Immunologic Activation. Peripheral blood was obtained via standard venipuncture methods. ‘Thespecimens were obtained between 9 and 11 A . V . i o control for circadian rhythm (21). Peripheral blood lymphocytes (PBL) were isolated by density-gradient centrifugation [22,23].The immunophenotypic: measure (HLA-DR’ cells) that was selected for use in this study has been shown to be a measure of immunologic activation [24]. Reagents were usecl as direct fluorescein isothiocyanate [FITC) or phy coerythrin (PE) conjugates at concentrations prcavi0u:;ly determined to be at antigen saturation for the number of cells studied 122, 231. All examinations were performed using d FACScari (Becton-Dickinson, Mt. View, CA) or a PROFILE (Coulter Electronics, Hialeah, FL) flow cytometer. These instruments were calibrated and stiindardized to produce equivalent data. Detailed information on the standardization of fluorescence intensity measurement and analysis have been p r ~iously :~ doscribed [22, 231. Percentage of HLA-DR positivity withiii the “lymphocyte” light scatter gate was adju stet1 based 011 CD45 and CD14 positivity in order to compensate for erythrocyte and monocyte contamination of this light scatter region. These corrected percentages were then used to determine the absolute number oi HLA-DR’ cells by multiplying the total lymphocyte count (obtained from a standard hematology profile) by the corrected HLA-DR percentage.
Data Analyses Although an extensive discussion of s,tnic.tural equation modeling is beyond the scope of this article [25, 261, a brief overview of the statistical methodology will be provided. Structural equation niotleling is a multivariate statistical technique that analyzes the direct contribution of one variable to ailother in a nonexperimental design [26]. Struc.tura1 equation modeling estimates the coefficient 01’ a set of linear structural equations that examine the hypothesized cause-effect relationships that are of interest to the
Vol. 5 , No. 4, December 1992
298 Parker et al.
Independent Variables
Dependent Variables
Disease Activity
TABLE 1
Means, Standard Deviations, and Pearson Intercorrelations for Variables in the Path Model BDI
Depression BDI Pain AH1
1.00
Pain 0.46*** 1.00
AH1
JC
0.26*
0.34** 0.18 0.29* * 1.00
0.12 1.00
Helplessness HLA-DR M SD
Pain
Activation
8.1 5.1
3.7 1.9
33.1 4.1
28.1 12.5
HLA-DR - 0.08
0.11 0.002 0.21* 1.oo
14.2 7.9
Note: B131, Beck Depression Inventory; pain, AIMS Pain Score; AHI, Arthritis Helplessness Index; JC, Joint Count; HLA-DR, percentage of cells in the peripheral blood that express HLA-DR. *p
Psychological Factors, Immunologic .Activation, and Disease Activity in Rheumatoid Arthritis Jerry C. Parker, Karen L. Smarr, Edgar 0. Angelone, Philip K. Mothersead, Byung S. Lee, Sara E. Walker, Alan 1. Bridges, and C. William Calclwell The purpose of this study was to use structural equation modeling techniques to examine potential interrelationships ctmong psychological factors, immunologic activation, and disease activity in rheumatoid arthritis (RA).The subjects were 80 male patients with a diagnosis of classic or definite RA. Measures included the Beck Depression Inventory, Ihe Arthritis Helplessness Index, and the Arthritis Impact Measurement Scales (AIMS) pain score. Joint counts and immunophenotypic analyses of periphrxal blood lymphocytes also were collected. Path cinalysis showed that percentage of HLA-DR cells in the peripheral Mood and helplessness were related to join count. In addition, joint count had an cffect upon depression. Depression had an effect upon pain. but there wus no reciprocal effect of pain upon depression. This study describes a preliminary path inodt:l of interrelationships among psychological fhctors, immunologic activation, and disease activ1 ty in RA. +
The interrelationships among psychological factors, Immunologic activation, and disease activity in rheumatoid arthritis (RA) constitute an area of conceptual
Jerrv C. Parker, PhI), Karen L. Smarr, BA, Edgar 0. Angelone, MA, Philip K. Mothersead. MA, Byung S. Lee, MA, Sara E. Walker, MD. Alan J. Bridges. !AD. and C. William Caldwell, MD, PhD, are with Harry S. Truman Memorial Veterans Hospital and University of Missouri-Columbia School of Medicine, Columbia, Missouri. Address correspondence to Jerry C. Parker, PhD, Psychology Service (116B),Harry :S. h u m a n Memorial Veterans' Hospital, 800 Hospital Drive, Colunibia, MO 65201. Submitted September 30, 1991; accepted March 31, 1992. 4 1992 by the Arthritis Hoalth Professions Association
196
and practical significance. Conceptually, there are important questions as to how RA disease activity may have an impact upon psychological states such as depression. Similarly, there are questions as to how R A disease activity relates to pain, because discrepancies between disease activity and selt-reported pain have been noted [1-3]. From a clinical perspective, an understanding of how symptoms such as depression, helplessness, and pain are manifested in RA is extremely important, but the methodological problems in this area of research are formidable. Several studies h a e found relationships between psychological variables and RA functional status [3-51, but studies using objective measures of disease activity are far less common. Even so, correlations between disease activity and psychological variables offer no information regarding causality. Psychological distress may exacerbate disease activity; but conversely, disease activity may exacerbate psychological distress. Reciprocal relationships are possible. The question of how psychological state and disease activity are related in RA has not been adequately answered by the correlational designs used to date. Eventually, data from prospective. randomized psychological intervention studies will be available, but an intermediate approach to exploring relationships between psychological state and RA disease activity involves structural equation modeling [6, 71. Structural equation modeling is a stiltistical technique that permits a researcher to postu1,ite an explanatory model and then to see if available data suppoirts the model. Structural equation modeling is a correlational technique, but involves a confirmatory (vs exploratory) statistical approach. Structural 0893-7'124192/$5 00
Psychological Factors 197
4rthritis Care and Research
equation modeling permits an analysis of how variables operate in combination. Therefore, structural equation modeling can play a particularly important role in the early stages of scientific inquiry. Hypothetical relationships can be examined in a preliminary way so thz t prospective, randomized studies can 3e designed more effectively. The purpose of this study was to use structural equation modeling techniques to examine potential interrelationships among psychological state, immunologic activ,ition, and disease activity in RA. Findings from prexious studies [3,8-121 contributed to the developnwnt of the proposed path model.
MATERIALS AND METHODS Subjects The subjects were 80 male patients from a midwestern Department ot Veterans Affairs hospital with a diagnosis of classic or definite RA. The diagnoses were made by a collaborating rheumatologist using the diagnostic criteria of the American Rheumatism Association [ 131 Exclusion criteria were as follows: (a) other uncontIolled medical disorders, (b) organic brain syndrome, [c) major psychiatric disturbance, (d) major communicative disorder, (e) history of severe noncompliance, (f) less than 7 years of education, and (g) illiteracy. The distribution of subjects by Steinbrocker Functional Class was as follows: Class I = 13%, Class I1 = 72%, and Class I11 = 15% [14]. The mean age for the sample was 61.6 (SD = 6.5) years. The mearl educational level was 11.0 (SD = 2.7) years. The mean Hollingshead index of socioeconomic status was 47.4. The mean disease duration for the sample was 13.4 years.
Measures Depression. The Beck Depression Inventory (BDI) was used to measure depression. The BDI is a 21item questionnaire that yields a single score [15].The BDI has good internal consistency, reliability, and validity [16]. Pain. The Pain Score from the Arthritis Impact Measurement Scales (AIMS) was used to measure pain [17]. The AIMS has been shown to be a reliable and valid instrument for measuring health status in RA patients [18 191. Helplessness. The Arthritis Helplessness Index (AHI) is a 15-item, self-report questionnaire that assesses patients’ perceptions of their ability to cope with
arthritis-related symptoms [lo].The AH1 yields a sin-gle score and has been shown to be a wliable and valid instrument [lo]. Disease Severity. Joint counts were used to measure disease activity. Joint counts were pix formed by an experienced rheumatology nurse clinician using the criteria of the American Rheumatism Association [ZO]. The procedure involved careful exarriination of each joint and a standardized quantification of painitenderness and swelling. Immunologic Activation. Peripheral blood was obtained via standard venipuncture methods. ‘Thespecimens were obtained between 9 and 11 A . V . i o control for circadian rhythm (21). Peripheral blood lymphocytes (PBL) were isolated by density-gradient centrifugation [22,23].The immunophenotypic: measure (HLA-DR’ cells) that was selected for use in this study has been shown to be a measure of immunologic activation [24]. Reagents were usecl as direct fluorescein isothiocyanate [FITC) or phy coerythrin (PE) conjugates at concentrations prcavi0u:;ly determined to be at antigen saturation for the number of cells studied 122, 231. All examinations were performed using d FACScari (Becton-Dickinson, Mt. View, CA) or a PROFILE (Coulter Electronics, Hialeah, FL) flow cytometer. These instruments were calibrated and stiindardized to produce equivalent data. Detailed information on the standardization of fluorescence intensity measurement and analysis have been p r ~iously :~ doscribed [22, 231. Percentage of HLA-DR positivity withiii the “lymphocyte” light scatter gate was adju stet1 based 011 CD45 and CD14 positivity in order to compensate for erythrocyte and monocyte contamination of this light scatter region. These corrected percentages were then used to determine the absolute number oi HLA-DR’ cells by multiplying the total lymphocyte count (obtained from a standard hematology profile) by the corrected HLA-DR percentage.
Data Analyses Although an extensive discussion of s,tnic.tural equation modeling is beyond the scope of this article [25, 261, a brief overview of the statistical methodology will be provided. Structural equation niotleling is a multivariate statistical technique that analyzes the direct contribution of one variable to ailother in a nonexperimental design [26]. Struc.tura1 equation modeling estimates the coefficient 01’ a set of linear structural equations that examine the hypothesized cause-effect relationships that are of interest to the
Vol. 5 , No. 4, December 1992
298 Parker et al.
Independent Variables
Dependent Variables
Disease Activity
TABLE 1
Means, Standard Deviations, and Pearson Intercorrelations for Variables in the Path Model BDI
Depression BDI Pain AH1
1.00
Pain 0.46*** 1.00
AH1
JC
0.26*
0.34** 0.18 0.29* * 1.00
0.12 1.00
Helplessness HLA-DR M SD
Pain
Activation
8.1 5.1
3.7 1.9
33.1 4.1
28.1 12.5
HLA-DR - 0.08
0.11 0.002 0.21* 1.oo
14.2 7.9
Note: B131, Beck Depression Inventory; pain, AIMS Pain Score; AHI, Arthritis Helplessness Index; JC, Joint Count; HLA-DR, percentage of cells in the peripheral blood that express HLA-DR. *p
