A randomized clinical trial of a psychoeducational intervention to improve outcomes in systemic lupus erythematosus
In a cross-sectional study, we previously identified 2 potentially modifiable risk factors for adverse outcomes in systemic lupus erythematosus (SLE): self-efficacy and social support. The goal of this study was to evaluate in a randomized controlled trial a theory-based intervention to improve patient self-efficacy and partner support to manage SLE.
Patients with SLE ages 18 years and older who met the American College of Rheumatology criteria and were able to identify a partner (spouse or family member) were recruited from 2 academic medical centers and randomized into an experimental group or a control group. Patients in the experimental group and their partners received an intervention designed to enhance self-efficacy, couples communication about lupus, social support, and problem solving, in the form of a 1-hour session with a nurse educator followed by monthly telephone counseling for 6 months. Patients in the control group and their partners received an attention placebo, including a 45-minute video presentation about lupus, and monthly telephone calls. Measures of physical and mental health status, disease activity, and psychosocial factors were collected at baseline, 6 months, and 12 months. The effect of the intervention on physical and mental health and disease activity at 6 and at 12 months was modeled with linear regression and adjusted for baseline health status, disease activity, sociodemographic factors, treatment change, and psychosocial factors.
One hundred twenty-two patients (plus their partners) were enrolled and randomized as follows: 64 to the experimental intervention and 58 to the attention control group. The participants were predominantly white, approximately half were college educated, and the groups were balanced for sociodemographic factors. At 6 months, significantly higher scores for couples communication (P = 0.01) and problem-focused coping (P = 0.03) were seen in the experimental group compared with the control group. At 12 months (6 months after the intervention ended), social support was higher (4.4 versus 4.1; P = 0.03), self-efficacy was higher (7.2 versus 6.2; P = 0.02), couples communication was higher (3.5 versus 3.1; P = 0.03), and fatigue was lower (5.1 versus 6.3; P = 0.02) in the experimental group compared with the control group. Global mental health status at 12 months, as measured by the Short Form 36 survey, was 69 points in the experimental group compared with 58 points in the control group (P = 0.04). In multivariate models, adjusting for baseline covariates, scores for couple communication (P = 0.01) were significantly higher at 6 months, and scores for self-efficacy (P = 0.004) and global mental health status (P = 0.03) were significantly higher at 12 months in the experimental group compared with the control group, and the mean score for global physical function was higher by 7 points, which was a clinically meaningful change (P = 0.2). The mean score for fatigue was also significantly lower in the experimental group than in the control group (P = 0.05). SLE disease activity was unchanged by this intervention.
This randomized, controlled trial of a theory-based educational intervention in SLE demonstrated significantly higher scores for couple communication, self-efficacy, and mental health status, and lower fatigue scores in the experimental group compared with the control group. Because couple communication and self-efficacy appear to be modifiable risk factors, they may also be potential targets in more disadvantaged populations.
Markers of an impoverished social environment such as low income, low educational attainment, public health insurance status, and nonwhite ethnicity predict worse morbidity and mortality across a broad range of medical conditions (1). In systemic lupus erythematosus (SLE), lower socioeconomic status is also an important factor in determining outcome (2–7). In 1997, we demonstrated that self-efficacy for disease management and social support were most strongly associated with disease activity and health status outcomes in SLE even when controlling for medical, behavioral, and various socioeconomic variables (8). What was more significant, perhaps, was that both self-efficacy and social support could likely be improved with targeted interventions, thus making these factors the first examples of modifiable, nonmedical risk factors associated with SLE outcomes. Since that time, other studies have confirmed the importance of psychosocial factors and outcome in SLE (9–15).
Meta-analyses show that a variety of psychoeducational interventions in rheumatic disease provide significant improvement in relief of pain, depression, and disability, over and above that of usual medical interventions (16, 17). Of 165 patient education studies in rheumatic diseases (17–19), only 9 specifically address SLE patients, and the results are inconsistent. Of these studies, 2 examined educational interventions designed to improve the health status of patients with SLE (20, 21), and although both demonstrated some benefit to patients, neither had a control group. A clinical trial of person-centered telephone counseling demonstrated significant improvements in psychological status but not in physical functioning in 15 patients with SLE (22). In another trial conducted by the same group of investigators, 58 patients with SLE were randomized to receive either telephone counseling or symptom monitoring, and significant improvement in physical function and social support was demonstrated in the counseling group (23). In contrast, a randomized clinical trial of brief supportive–expressive group psychotherapy in 133 patients with SLE from 9 clinics across Canada showed no significant benefits in psychological distress, medical symptoms, health care costs, and quality of life (24).
Although psychoeducational programs significantly improve outcome in rheumatic disease (25, 26), it is less evident which kinds of programs work best, for which kinds of patients, and in which situations. DeVellis and Blalock (26) note that the diverse causes of disability require a repertoire of approaches if they are to be effective, and that theory-based patient education appears to work better than empiric approaches. Therefore, we undertook to test a theory-based psychoeducational intervention to improve patient selfefficacy, partner support, and problem-solving skills of the patient–partner pair to manage SLE in a randomized, controlled trial. In addition, we wanted to test whether such an intervention could improve patient self-management behaviors and health status, and reduce SLE disease activity, especially fatigue, which is perhaps the most common symptom.
SUBJECTS AND METHODS
Subjects and entry criteria.
The study group comprised patients with SLE according to the American College of Rheumatology revised criteria (27), who had disease of any duration, were age ≥18 years, and whose physicians consented to the recruitment of their patients. Subjects were selected by random sampling of SLE registries for patients who had visited a rheumatologist at least twice in the prior year at the Brigham and Women's Hospital and the Beth Israel Deaconness Hospital. Patients were recruited by mail. Subjects meeting the above criteria also had to identify a partner (spouse or family member) who was willing to participate in the study. The study was approved by the Institutional Review Boards at both hospitals.
We pilot tested a counseling intervention for one session on 18 SLE patients and their partners. At baseline, 12 patients had at least one area of very low efficacy for self-management (score of ≤5 on a 0–10-point scale), the most common being management of fatigue. Fifteen couples (75%) had at least one area of significant disagreement, defined as a difference of ≥3 points. The efficacy survey for patients and partners was repeated at 2 weeks. All patients, and all but one partner, reported benefits from the discussion—even couples with no initial self-management problems or disagreements in their answers to questions about efficacy. Two-thirds of couples reported changes (all positive) in their coping, supportive behavior, or communication as a result of the session. Based on the results of this pilot study, our final study procedures, intervention, and outcome measurements were refined prior to beginning the randomized controlled trial.
Recruitment and randomization procedures.
Patients were mailed a letter describing the study. Patients who, at the time of a followup telephone call, agreed to participate were provisionally enrolled and asked to select a partner for the study. Preferred partners were spouses who lived with the patient and/or nonspouses who were in a position to provide support to the patient in lupus management (28). Consent forms and baseline questionnaires were mailed to patients and their partners. They were asked to fill these out independently and mail them back separately, and not to share this information with each other until the time of the educational session.
Each patient who returned the baseline questionnaire and provided informed consent was placed into 1 of 4 randomization groups, based on his or her baseline questionnaire scores for overall self-efficacy and mental health (high–high, high–low, low–high, low–low). Mental health was used as a randomization stratum because significant differences in health care utilization have been associated with mental health status among patients with chronic disease (29, 30). Efficacy was used as a randomization stratifier, because it is a key intervention-mediating variable. Cutoff points for strata were based on the median scores of patients participating in the pilot study. Within each stratum, patients and their partners were randomly assigned, as pairs, to either the experimental or control group to balance the study groups on key elements thought to affect intervention impact. Each stratum had its own randomization list, with balanced blocks, to ensure balance over time within strata; this is an important consideration, because of the possibility of systematic change in counselor behaviors (e.g., growth in skill) over the course of the study.
Theoretical framework for the intervention.
The intervention builds on major theory described by Bandura (31, 32) and Ewart (33). Perceived self-efficacy, or confidence in one's ability to carry out behaviors with desired consequences, was conceptualized by Bandura (31) as a key integrating concept in cognition and behavior. In patients with rheumatic illness, Holman and Lorig (34, 35) demonstrated that perceived self-efficacy to cope with the consequences of chronic disease is essential to developing self-management capabilities, and that perceived self-efficacy can be enhanced by appropriate learning experiences (32, 36). Ewart's framework for applying social psychological approaches to the design of health interventions (33) emphasizes social interdependence and interaction in personal control of health behavior, and a key integrating concept in this model is behavioral self-efficacy. Spousal perceptions of a patient's self-efficacy can be supportive or nonsupportive of patient health behaviors (37). An intervention in which wives participated in their husbands' cardiac rehabilitation brought their appraisals of the husbands' efficacy and the husbands' self-appraisals much closer, presumably with beneficial effects on wives' supportiveness for their husbands' participation in physical activity (38, 39).
Supportive behaviors are emotional (listening, providing empathy), instrumental (providing resources such as babysitting, rides, money), and informational (providing information about means and resources for self-management) (40). In coping with a chronic disease, naturally occurring support, such as that from a friend or family member, may be more potent than contrived support (26), but telephone support from nurses or trained research assistants improves health in arthritis and other chronic disease (29, 41–43).
Although theorists have tended to emphasize individuals as the unit of analysis, a social–contextual view recognizes that success in many health areas depends on close personal relationships and on a partner's ability to collaborate effectively in problem solving (33), and thus we used the problem-solving approach described by D'Zurilla and Goldfried (44). Chronic diseases pose or exacerbate numerous problems of living for those afflicted. Problem solving is one strategy of coping that is more useful than many other strategies, such as wishful thinking or blaming others for one's problems. Researchers have examined the potential for teaching basic problem-solving skills to persons with chronic disease, including rheumatoid arthritis and SLE (22, 29, 45). However, this approach has not been widely tested.
The intervention was carried out during a 30–45-minute discussion between educator, patient, and partner, after a regular visit for medical care. Based on the literature, we decided that the most effective (and acceptable) intervention would be one “owned” by the patient. Hence, patients chose their own partners, the pair identified the problems they wished to work on, and they came up with their own plans, as summarized below.
Before the session, the patient and partner independently completed efficacy questionnaires, the patient for himself or herself, and the partner for the patient, covering important lupus self-management behaviors, such as management of pain, fatigue, depression, appointment-keeping, and adherence to medications. The counselor then met with the subjects and shared their responses. The first aim was to identify areas in which patient self-efficacy for lupus management was low, with the goal of helping the patient and partner each identify ways in which to improve the patient's self-management abilities. The couple was introduced to a problem-solving approach, including problem identification, generation and evaluation of alternative solutions, choosing a solution, and making concrete plans. Couples were encouraged to generate their own solutions, based on past successful behaviors, and the counselor served as a facilitator. The second aim was to identify discordant patient and partner beliefs about patient self-management abilities, to help them come to a common understanding of the patient's actual abilities and improve their communication about lupus management. Solutions included techniques such as monitoring (e.g., having the partner regularly observe the patient for signs of fatigue) and checking (e.g., having the partner ask the patient whether he/she is fatigued). The focus was lupus management and not generic improvement of the pair's communication.
A brief written summary of the meeting was provided to the patients' doctors in order to alert them to relevant issues, such as patient questions or misconceptions, difficulty following the regimen, and need for encouragement. This integrated the educational session into the patient's medical care, as would be natural and desirable in any team care setting.
The intervention was followed by 5 telephone calls. The first telephone call included the patient and partner, if possible. Subsequent calls were made to the patient each month for 4 more months, and included the partner, if both the patient and partner agreed. Collectively, the calls served to monitor and encourage progress in problem solving and communication and to help the patient and partner to renegotiate these goals as their experience in self-management and communication grew. At each call, a brief screening instrument for SLE activity, the Systemic Lupus Activity Questionnaire (SLAQ) (46), was administered. Each telephone call lasted 10–25 minutes.
The intervention was delivered by a masters-level nurse (HE) with >20 years of clinical experience, who was trained in proven efficacy-enhancing counseling techniques. For quality control, audiotapes of the counseling sessions were periodically reviewed and discussed with the supervisor (LHD). A psychiatrist (MPR) provided consultation when the nurse had a question about the couple's need for more therapeutic intervention.
Patients and their partners received an attention intervention, which was a 45-minute meeting with the counselor during which they viewed a videotape of a lupus patient talking about what lupus was like for her (47). Subjects were able to ask questions about the film, were given pamphlets about lupus and its management, and were encouraged to query their doctors. The counselor did not discuss self-efficacy or patient–partner communication, or address specific problems in lupus disease management.
At 1 month, control group patients and their partners were telephoned to determine whether they had any questions about the information they received at the initial visit. Subsequent monthly telephone calls involved the patient only and were devoted to collection of SLE activity information (the SLAQ). However, there was no discussion of self-management. These telephone calls lasted 3–5 minutes.
The intent of the intervention was not to be a proxy for psychological care or to solve fundamental or pathologic deficiencies in life skills or couples communication patterns, but to provide information about practical and specific problem-solving approaches and behaviors that impact lupus management. Patients who were identified, at any point, as being in need of psychiatric, therapeutic, or couples counseling were advised to seek professional help after consultation with a psychiatrist (MPR). Similarly, patients in whom emergent medical needs were identified were referred immediately.
Data collected included age, sex, ethnicity, education, work status, occupation, income, insurance, and marital status. The primary general health status study outcome was measured by the Medical Outcomes Study Short Form 36 (SF-36) questionnaire (48, 49), which has 2 major domains—physical health and mental health—and measures function, pain, work and social roles, vitality, and mental health. The SF-36 has been used extensively in health outcomes studies (48) in the general population and in SLE patients and has excellent reliability (the subscales have internal reliability of 0.85 or better) and validity. Disease-specific outcome was measured with the SLAQ, a validated measure of self-reported SLE activity that correlates well with a physician- rated disease activity measure, the Systemic Lupus Activity Measure (SLAM) (46). Fatigue is a sensation of weariness or tiredness that is extremely common and may be the rate-limiting symptom in SLE (50, 51) and is represented in both the SF-36 (vitality subscale) and the SLAQ. However, these instruments do not provide precise enough measures of this important variable to track it reliably. Fatigue was assessed by a scale designed specifically for lupus patients (50), with demonstrated reliability (α = 0.89) and responsiveness to an aerobic conditioning treatment program used in a clinical trial in SLE patients (50,51).
The following measures track the pathways by which the intervention was expected to affect health status. Self-efficacy, which is an important predictor of patient behavior, responds to efficacy-enhancing interventions (49), mediates behavior change, and explains much of the success of the Arthritis Foundation's arthritis self-help programs (34). We used the self-efficacy–other summated scale described by Lorig et al (50), adapted for lupus, with an internal reliability of α = 0.91 in our studies. Social support was measured with a modified social support scale developed for arthritis patients (40). This reliable scale (Cronbach's α = 0.84) assesses instrumental, emotional, and informational support. In our research (8) it is significantly associated with health status in lupus patients. Problem solving was measured by an 11-item coping measure, adapted from the method described by Stone and Neale (51) for rheumatic disease patients (52). This measure assesses the frequency of use of problem-focused and emotion-focused coping strategies in managing common problems of rheumatic disease. Satisfaction with medical care was assessed with the 8-item subscale of the Medical Interview Satisfaction Scale (53), which has a high internal reliability (α = 0.89), and focuses on satisfaction with communication with the doctor. Compliance with prescribed medical treatment was assessed using a 4-item measure based on a questionnaire designed by Morisky et al (54), with acceptable internal reliability (α = 0.65) and ability to predict both physiologic and health outcome (28). It assessed patterns of missed or extra units of treatment and reasons for deviation from prescription and compliance with visits to the physician. Using this information in our previous research, we have been able to judge patients as being compliant or noncompliant, with 87% agreement between 2 blinded raters and a kappa value of 0.74 for interrater reliability. Despite the well-known limitations of self-reported compliance, such reports are highly specific for noncompliance (55) and have been useful predictors of health outcome in reports of chronic disease (28, 56).
Models were adjusted for sociodemographic factors, age, education level, ethnicity, and treatment change. Treatment change during the 12-month trial was assessed by a single question regarding whether the physician had increased the prednisone dosage since the time of the last followup. Responses were scored as a variable for any prednisone dosage increase within 3 months of baseline, any prednisone dosage increase within 6 months of baseline, and any prednisone dosage increase within 12 months of baseline.
Sociodemographic characteristics of the experimental group were compared with those of the control group by Student's t-tests. The primary outcomes were defined as follows: global physical function was defined as the average score on 3 subscales of the SF-36 (physical functioning, role limitations due to physical functioning, and pain). Global mental health status was defined as the average score on 3 subscales of the SF-36 (social functioning, role limitations due to emotional problems, and mental health). This scoring system was used in the original multicenter study of modifiable risk factors for SLE (8) rather than using the physical component score (PCS) and the mental health component score (MCS). The PCS/MCS dimensions are based on regression equations so that each one of 8 subscales contributes to the score, and regression coefficients are derived from the factor analysis of the dimensions with a negative loading yielding a negative coefficient. For example, in 2 subjects with identical physical activity subscale scores, 1 subject would have a lower PCS if he or she had a depressed score on the mental health subscale, because the mental health dimension loads negatively in the physical dimension. Disease activity was defined as the total patient self-reported SLAQ score based on individual symptoms, weighted in a manner analogous to that for the SLAM, a physician-administered disease activity measure (46). Fatigue, defined as a sensation of weariness or tiredness that is extremely common and may be the rate-limiting symptom in SLE (57, 58), was measured by a 4-item lupus fatigue scale, which has good reliability and validity (50,51).
The 5 primary outcomes described above were measured at baseline and at 6 and 12 months. The effect of the intervention on the 5 major outcome variables, with separate models for each outcome at 6 months and each outcome at 12 months, was analyzed in an intent-to-treat analysis, with linear regression models controlling for baseline scores of outcome variables as well as other covariates of interest. All models started with all the covariates of interest, which included baseline values of positive social support, negative social support, self-efficacy, satisfaction, compliance, couples communication, emotion-focused coping, problem-focused coping, fatigue, patient–partner agreement on self-efficacy, SF-36 global mental health, SF-36 global physical function, and problem solving, as well as demographic variables including age, education, and ethnicity. Six-month outcome models were additionally adjusted for any prednisone dosage increase within 3 months of baseline and any increase in the prednisone dosage within 6 months of baseline. Twelve-month outcome models were additionally adjusted for any increase in the prednisone dosage within 12 months of baseline. Stepwise regression was used to reduce the original models to those models with only significant main factors. Interactions between the intervention group and the significant main factors were then explored.
The effectiveness of the intervention was expected to operate through intervening variables, such as self-efficacy and social support, that are not the ultimate outcome of interest. We used linear regression models to predict these intermediate outcomes, adjusting for baseline values of covariates. Stepwise regression was used to reduce the original models to those models with only significant main factors. Interactions between the intervention group and the significant main factors were then explored. In a secondary analysis, we tested the effect of forcing 6-month psychosocial variables into the stepwise models predicting outcomes at 12 months.
Of 334 eligible patients contacted, 150 (45%) consented to participate; of these, 14 patients withdrew prior to randomization, and 14 were eliminated due to lack of partner. Sixty-four SLE patients (plus their partners) were randomized to the experimental group, and 58 SLE patients (plus their partners) were randomized to the control group. Of the 122 SLE patient–partner pairs randomized, 15 withdrew before the intervention; 107 patients returned the baseline questionnaire and attended at least the first visit with the nurse educator. An intent-to treat analysis was performed on all 122 randomized patients. At 6 months, 94 patient–partner pairs (77%) completed the followup questionnaires by telephone. At 12 months, 90 patient–partner pairs (74%) completed the followup questionnaires by telephone. The participants were predominantly white and college educated, and the groups were balanced for sociodemographic factors (Table 1). The mean baseline global mental health score of 61 for women with SLE in this study is 18 points lower than reported norms for women ages 35–44 years (48).
Table 1. Characteristics of the study patients
|Mean ± SD age, years||42.7 ± 22.8||40.8 ± 11.1|
|No. (%) white||57 (89)||48 (84)|
|No. (%) female||63 (98)||56 (98)|
|No. (%) with college education||47 (73)||39 (68)|
|No. (%) below national poverty level||6 (9)||4 (7)|
Univariate analysis of intermediate outcomes demonstrated that at 6 months, results for couples communication and problem-focused coping were significantly higher in the experimental group than in the control group (Table 2). At 12 months, positive social support and patient self-efficacy scores were significantly higher and the fatigue score was significantly lower in the experimental versus the control group. For the primary outcomes, disease activity and health status at 12 months, global mental health status was significantly higher in the experimental group than in the control group, and global physical function showed a modest trend for improvement in the experimental group compared with the control group (Table 3). Disease activity was improved at 6 months in both the experimental and control groups, and was not significantly different between groups at 12 months.
Table 2. Univariate analysis of intermediate outcomes at baseline, 6 months, and 12 months*
|Social support, negative||2.7 ± 0.9||2.6 ± 0.9||0.6||2.7 ± 0.9||2.7 ± 0.8||0.8||2.7 ± 0.9||2.7 ± 0.9||0.9|
|Social support, positive||4.1 ± 0.8||4.1 ± 0.7||0.8||4.3 ± 0.6||4.1 ± 0.7||0.17||4.4 ± 0.6||4.1 ± 0.6||0.03|
|Self-efficacy, patient||6.3 ± 2.3||6.2 ± 2.0||0.9||6.9 ± 2.1||6.2 ± 2.0||0.1||7.2 ± 1.9||6.2 ± 2.0||0.02|
|Couples communication||3.2 ± 0.9||3.1 ± 0.8||0.5||3.4 ± 0.8||2.9 ± 0.7||0.01||3.5||3.1||0.03|
|Emotion-focused coping||2.6 ± 0.8||2.4 ± 0.8||0.3||2.5 ± 0.9||2.4 ± 0.7||0.5||2.6 ± 0.8||2.5 ± 0.9||0.8|
|Problem-focused coping||2.7 ± 0.8||2.6 ± 0.7||0.7||2.8 ± 0.8||2.5 ± 0.7||0.03||2.8 ± 0.7||2.6 ± 0.7||0.3|
Table 3. Univariate analysis of disease activity and health status outcomes at baseline, 6 months, and 12 months*
|SLAQ score||13.3 ± 6.0||14.4 ± 7.1||0.4||8.5 ± 6.6||10.8 ± 7.5||0.1||10.7 ± 7.6||11.4 ± 6.7||0.6|
|SF-36, global mental health||61 ± 23||61 ± 23||1.0||61 ± 25||61 ± 25||1.0||69 ± 26||58 ± 23||0.04|
|SF-36, global physical function||48 ± 24||49 ± 25||0.9||52 ± 26||47 ± 24||0.4||55 ± 25||48 ± 25||0.2|
|Fatigue||5.8 ± 2.2||6.4 ± 2.2||0.09||5.3 ± 2.5||6.1 ± 2.3||0.1||5.1 ± 2.4||6.3 ± 2.4||0.02|
Multivariate models demonstrated that disease activity and health status outcomes were associated with the baseline values for these variables. Adjusting for baseline covariates, including sociodemographic and psychosocial variables, and baseline values for the main outcomes, we demonstrated that improved physical function at 6 and 12 months was associated with white ethnicity (P = 0.008), no increase in the prednisone dosage within 12 months (P = 0.02), and younger age (P = 0.03) (Table 4). Improved mental health status at 6 months was associated with higher baseline mental health status (P = 0.0001) and less fatigue at baseline (P = 0.002). Improved mental health status at 12 months was associated with higher baseline mental health status (P = 0.0004) and higher self-efficacy at baseline (P = 0.001) as well as assignment to the experimental group (P = 0.03). Assignment to the experimental group did not significantly predict physical function or disease activity at either 6 or 12 months.
Table 4. Multivariate linear regression models for predictors of disease activity, health status outcomes, and fatigue at 6 months and 12 months*
|Less disease activity at 6 months|| || || |
| Less disease activity at baseline||0.0001||0.1||0.7|
| Lower satisfaction at baseline||0.04||0.8||1.6|
|Less disease activity at 12 months†|| || || |
| Less disease activity at baseline||0.0001||0.1||0.8|
|Improved physical function at 6 months|| || || |
| Higher physical function at baseline||0.0001||0.1||0.7|
| White ethnicity||0.002||6.2||19.6|
|Improved physical function at 12 months|| || || |
| Higher physical function at baseline||0.0001||0.1||0.7|
| White ethnicity||0.008||5.8||15.6|
| Prednisone dosage increase||0.02||3.5||−8.2|
|Improved mental health status at 6 months|| || || |
| Higher mental health at baseline||0.0001||0.1||0.5|
| Less fatigue at baseline||0.002||1.0||−3.3|
|Improved mental health status at 12 months|| || || |
| Higher mental health at baseline||0.0004||0.1||0.4|
| Higher self-efficacy at baseline||0.001||1.2||4.1|
| Experimental group assignment||0.03||4.0||9.0|
|Less fatigue at 6 months|| || || |
| Less fatigue at baseline||0.001||0.1||0.8|
|Less fatigue at 12 months|| || || |
| Less fatigue at baseline||0.0001||0.1||0.6|
| Higher self-efficacy at baseline||0.02||0.1||0.3|
| Experimental group assignment||0.05||0.4||0.7|
We measured change in intermediate variables in order to model the pathways by which the intervention has its effects. Multivariate analysis of intermediate psychosocial outcomes at 6 and 12 months demonstrated significant relationships between baseline values of the psychosocial factor and its corresponding outcome variable (Table 5). For example, baseline self-efficacy was predictive of self-efficacy at 6 months and at 12 months. Assignment to the experimental group predicted better 6-month outcomes for self-efficacy, problem-focused coping, and couples communication, adjusting for demographic, psychosocial factors, and baseline global physical function. In addition, assignment to the experimental group predicted better 12-month outcome for both self-efficacy and fatigue. We found no significant interactions between group assignment and baseline psychosocial factors and psychosocial outcomes.
Table 5. Multivariate linear regression models for intermediate outcomes at 6 months and 12 months*
|Higher social support at 6 months|| || || |
| Higher social support at baseline||0.0001||0.1||0.5|
| Higher coping with emotions at baseline||0.03||0.1||0.2|
|Higher social support at 12 months†|| || || |
| Higher social support at baseline||0.0001||0.1||0.8|
|Higher self-efficacy at 6 months|| || || |
| Higher self-efficacy at baseline||0.0001||0.1||0.7|
| Higher emotion-focused coping at baseline||0.03||0.2||0.4|
| Experimental group assignment||0.04||0.3||0.5|
|Higher self-efficacy at 12 months|| || || |
| Higher self-efficacy at baseline||0.0001||0.1||0.6|
| Higher physical function at baseline||0.03||0.01||0.02|
| Experimental group assignment||0.004||0.3||0.9|
|Higher couple communication at 6 months|| || || |
| Better couple communication at baseline||0.0001||0.1||0.5|
| Higher emotion-focused coping at baseline||0.009||0.1||0.2|
| Experimental group assignment||0.01||0.1||0.3|
|Higher couple communication at 12 months|| || || |
| Higher couple communication at baseline||0.0001||0.1||0.7|
| Higher emotion-focused coping at baseline||0.02||0.1||0.2|
|Higher emotion-focused coping at 6 months|| || || |
| Higher emotion-focused coping at baseline||0.0001||0.1||0.6|
| White ethnicity||0.02||0.2||−0.4|
| Better coping with problems at baseline||0.03||0.1||0.2|
|Higher emotion-focused coping at 12 months|| || || |
| Higher emotion-focused coping at baseline||0.0001||0.08||0.8|
| Visit compliance at baseline||0.04||0.08||−0.2|
| Higher education at baseline||0.03||0.13||0.3|
|Higher problem-focused coping at 6 months|| || || |
| Higher problem-focused coping at baseline||0.0001||0.1||0.7|
| Medication compliance at baseline||0.03||0.1||−0.2|
| Experimental group assignment||0.05||0.1||0.2|
|Higher problem-focused coping at 12 months|| || || |
| Higher problem-focused coping at baseline||0.0005||0.1||0.3|
| Higher emotion-focused coping at baseline||0.0005||0.1||0.3|
| Medication compliance at baseline||0.02||0.2||−0.2|
The small number of subjects did not support analytic modeling, but we did examine the effect of forcing 6-month psychosocial variables into the models predicting mental health at 12 months. Self-efficacy and emotion-based coping were both significant predictors of 12-month mental health, but when baseline variables were added back, only baseline and 6-month self-efficacy predicted 12-month mental health; assignment to the experimental group was not significant (P = 0.32).
This randomized, controlled trial demonstrated that a theory-based psychoeducational intervention to improve patient self-efficacy, partner support, and problem solving for the management of lupus could improve couples communication and problem-focused coping by the end of the 6-month intervention. The intervention resulted in significant improvement in patient self-efficacy at both 6 months and 12 months after the first counseling session (6 months after the last telephone contact) and significant improvement in fatigue at 12 months. At the 12-month followup, the intervention resulted in significant improvement in quality of life on the SF-36, in terms of global mental health status, and significant improvement in fatigue, after adjusting for baseline psychosocial, demographic, and clinical factors. Global physical function was improved by 7 points on the SF-36; although this change was not statistically significant, it is greater than the minimum clinically significant change suggested for interpreting change scores in rheumatoid arthritis trials (59). The clinical trial results confirm cross-sectional findings from research done by our group and other investigators, and support the importance of psychosocial factors in outcome in SLE (8–15).
Self-efficacy has been correlated with SLE outcome in several studies (8–15) and has been found to consistently predict health-related outcomes in prospective studies (60). In this trial, we demonstrated that self-efficacy could be enhanced by our intervention and that it was a key predictor of mental health status and fatigue outcomes. These results are consistent with results of another randomized, controlled trial of counseling for SLE patients in a group setting, which demonstrated significant improvements in psychological status but not in physical functioning and social support (23). That trial did not involve a partner but was similar to our trial, in that the counselor targeted 6 patient behaviors: self-care activities in managing fatigue, patient's communication skills, removing barriers to medical care, medication self-management, symptom monitoring, and stress control methods. Interestingly, in that intervention, the primary mediating variable strongly associated with health outcomes was fatigue self-efficacy.
Social support is associated with better health through multiple routes (61, 62), including physiologic buffering against stress and improved coping. Randomized trials of teaching problem-solving skills have shown some promise in rheumatic diseases (29, 45, 63, 64). In this trial, our finding that 99 (81%) of 122 couples had significant differences in their estimates of patient efficacy in coping with the disease provides empiric support for our couples-based counseling model. We demonstrated that a targeted intervention in SLE that involved a supportive partner resulted in improvement in couples communication, problem-focused coping, and social support. However, these factors were not independently associated with the main outcomes of health status and disease activity. We also found no evidence for interaction between baseline levels of these factors and the intervention, which suggests that the intervention is beneficial at all levels of self-efficacy, coping skills, and social support.
Some limitations of the study may reduce generalizability of the results. Subjects were recruited from 2 teaching hospitals and may not be representative of SLE patients in general. In spite of the random sampling methods, there is a possibility of self-selection bias; only persons who were sufficiently motivated to respond to a mailed questionnaire were included. Not surprisingly, the sample comprised predominantly white and college-educated patients. A secondary analysis of our previous study (8) of 200 patients, balanced for ethnicity and socioeconomic status, indicated that social support was most beneficial for patients whose disease was mild, whose income was greater than poverty level, and whose ethnicity was white and educational level was higher (65). This describes our current study population well, and it is possible that our intervention may be less effective in a more socially disadvantaged population. In addition, not all participants could identify a partner. However, many patients with lupus end up in tertiary care centers, and these are precisely the patients who might benefit most from interventions such as ours. We used a single educational counselor, which may limit the generalizability of our results. However, we note that many interventions have used only a single counselor, yet efficacy results are generally similar across interventions (16).
Our intervention, as noted above, combined elements of efficacy enhancement, problem solving, and social support enhancement. With our design, we cannot definitively separate the effects of each element. However, we wanted the intervention as a whole to be stronger than individual elements. Our secondary analysis indicates that the intervention worked at least through manipulation of patient self-efficacy.
In summary, our theory-based patient education intervention combined elements of efficacy enhancement, social support, and problem solving in a novel and effective manner. Patients with SLE had a modest and significant improvement in mental health status and fatigue despite persistent disease activity. This nurse-delivered intervention has the potential to reduce health problems and costs in a debilitating, management-intensive chronic disease. Cost data will be analyzed in a separate report.
We thank the lupus patients and their partners who participated in this trial and acknowledge the help of Charlotte Phillips, RN, MPH, with performing the interventions.