|Home | About | Journals | Submit | Contact Us | Français|
A pretest–posttest, repeated-measures design was used to evaluate the effects of two stress management interventions on a battery of outcomes derived from a psychoneuroimmunological (PNI) framework. The effects of cognitive-behavioral relaxation training groups (CBSM) and social support groups (SSG) were compared with a WAIT-listed control group on the outcomes of psychosocial functioning, quality of life, neuroendocrine mediation, and somatic health. Participants were 148 individuals (119 men, 29 women), diagnosed with HIV disease; 112 (76%) completing the study groups. Using analysis of covariance, the CBSM group was found to have significantly higher postintervention emotional well-being and total quality-of-life scores than did either the SSG or WAIT groups. SSG participants had significantly lower social/family well-being scores immediately postintervention and lower social support scores after 6 months. The findings point to a pressing need for further, well-controlled research with these common intervention modalities.
Although it remains potentially fatal, infection with the human immunodeficiency virus (HIV) has become eminently more treatable as a chronic illness with the advent of highly active antiretroviral therapies (HAART) that include protease inhibitors (Carpenter et al., 1998; Centers for Disease Control and Prevention, 1998; Powderly, Landay, & Lederman, 1998). However, people with HIV disease continue to experience significant stressors in living with this potentially catastrophic chronic illness. Stressful disease-management issues accompany the ongoing psychological stressors associated with such factors as stigmatization, illness-related uncertainty, and existential issues related to diagnosis with a life-threatening chronic illness (Folkman, 1993; Kelly, 1998; Lutgendorf, Antoni, Schneiderman, Ironson, & Fletcher, 1995; Weiss, 1997; Wilson, Hutchinson, & Holzemer, 1997). In studies with sufficient methodological rigor, evidence of a negative impact of psychological distress on HIV disease progression is clearly emerging (Cole & Kemeny, 1997; McCain & Zeller, 1996; Robinson, Mathews, & Witek-Janusek, 1999). This randomized clinical trial was designed to test the effects of two stress management interventions on indicators of psychosocial functioning, quality of life, neuroendocrine mediation of stress, and somatic health in persons with HIV disease.
Insights as to the relationship of psychological and physiological health in HIVand other diseases are emanating from research in psychoneuroimmunology (PNI). The PNI paradigm accounts for the negative impact of perceived stress on HIV disease progression, primarily as a function of immunosuppression mediated by elevated cortisol. Thus, chronic and/or severe psychological stress associated with living with HIV disease may further compromise immune functioning over the illness trajectory. The theoretical framework for this research integrated Lazarus and Folkman’s (1984) cognitive-transactional model of the stress process within the PNI paradigm. This stress process model holds that the variety of coping strategies (including relaxation and social support) used in response to perceived stress serve to psychologically mediate the adaptational outcomes of psychosocial functioning, quality of life, and somatic health. The PNI framework synthesizes psychosocial, neuroendocrine, virological, immunological, and somatic health constructs within the context of HIV disease. The conceptual model for the study, indicating study constructs and concepts with the selected indicators and corresponding measures, is presented as Figure 1.
A growing body of research with persons who have HIV disease, as well as those who have other chronic and potentially fatal illnesses such as cancer, indicates that not only can a variety of biobehavioral strategies for stress management mitigate psychological distress and improve coping skills, they also can enhance immune function through neuroendocrine-immune system modulation (Cole & Kemeny, 1997; Glaser & Kiecolt-Glaser, 1994; Schulz & Schulz, 1992). Such intervention strategies thus have considerable potential for attenuating both psychological and physical symptomatology, thereby improving the quality of life of persons with HIV disease.
A series of intervention studies conducted at the University of Miami Center for the Biopsychosocial Study of AIDS documented negative correlations between natural killer (NK) cell cytotoxicity and both anxiety and avoidant thinking, suggesting that psychological distress related to a seropositive diagnosis may be associated with decreased NK cell activity (Antoni, Schneiderman, et al., 1991; Ironson et al., 1990). The Miami group reported that asymptomatic individuals who had received a 10-week cognitive-behavioral stress management (CBSM) intervention (involving relaxation and coping skills training) had less anxiety and depression, increased CD3+/CD4+ (helper T-lymphocyte) counts, and higher NK cell counts than did individuals in control groups (Antoni, Baggett, et al., 1991; Esterling et al., 1992; LaPerriere et al., 1991). More recently, the Miami group conducted a randomized clinical trial (Lutgendorf et al., 1998) of a 10-week CBSM intervention that included emphases on cognitive and behavioral coping skills as well as social support in a group of symptomatic HIV-seropositive men. In comparison with the wait-listed control group (n =18), the CBSM group (n =22) demonstrated enhanced coping skills involving cognitive restructuring and acceptance and enhanced perceptions of social support, both of which were shown by regression analyses to mediate improved psychological well-being.
More recent work has continued to support the use of CBSM as an effective strategy in the management of distress associated with HIV disease. Cruess et al. (1999) found that a 10-week CBSM intervention buffered decreases in plasma dehydroepiandrosterone-sulfate (DHEA-S) and increases in the cortisol:DHEA-S ratio in 43 HIV-seropositive men. Changes in the cortisol:D-HEA-S ratio were significantly and positively related to changes in total mood disturbance and perceived stress over time. In a similar study of 30 HIV-seropositive men who participated in a 10-week group-based CBSM intervention, Cruess, Antoni, Kumar, and Schneiderman (2000) found that presession salivary cortisol levels decreased across the 10-week period and were related to decreases in global measures of total mood disturbance and anxiety. Further, reductions in presession cortisol levels were associated with decreases in self-reported stress levels during home practice of relaxation techniques. Antoni, Cruess, S., et al. (2000) tested the effects of a 10-week CBSM intervention on 24-hr urinary free-cortisol levels and distressed mood in 40 symptomatic HIV-seropositive men. These researchers found significantly lower post-treatment levels of self-reported depressed affect, anxiety, anger, and confusion in the intervention group as compared with those in a wait-listed control group (n =19). Participants assigned to the intervention group also had significantly less cortisol output in comparison with the control group.
Antoni, Cruess, D.G., et al. (2000) further reported the effects of a 10-week CBSM intervention on anxious mood, stress, 24-hr urinary catecholamine levels, and changes in T-lymphocyte subpopulations over time. In 47 HIV-infected gay men who participated in the intervention, the researchers reported significantly lower post-treatment levels of self-reported anxiety, anger, total mood disturbance, and perceived stress and lower norepinephrine (NE) output compared with men in a wait-listed control group (n =26). Anxiety decreases paralleled NE reductions at an individual level, and greater decreases in NE output and a greater frequency of relaxation home practice during the intervention period predicted higher CD3+/CD8+ (T-cytotoxic/suppressor lymphocyte) cell counts at follow-up. Significantly greater numbers of T-cytotoxic/suppressor lymphocytes also were found 6–12 months later in the individuals assigned to the CBSM intervention group as compared with the wait-listed group.
Although findings concerning the psychological stress-moderating effects of social support have generally been positive, results from studies with immunological variables have been inconsistent. Evans, Leserman, Perkins, Murphy, & Folds (1992) reported stress-moderating effects of social support and a positive relationship between the number of supportive relationships and the percentage of CD4+ T cells among 63 HIV-infected men. However, in two larger studies (Perry, Fishman, Jacobsberg, & Frances, 1992; Rabkin et al., 1991), no significant relationships were found between CD4+ lymphocyte levels and either social support or stress.
In one of the few well-controlled experimental studies of therapeutic social support, Kelly et al. (1993) compared 8-week group interventions of cognitive-behavioral or social support therapy. Participants were 68 men at various stages of HIV disease who were classified as depressed and had requested mental health services. In comparison with the no-treatment group, both therapy groups had less postintervention depression, hostility, and somatization. Maintenance of these clinical improvements after 3 months was higher for the social support group participants than for the cognitive-behavioral and control groups, but there was no report of longer-term participant follow-up. For ethical reasons with this depressed sample, persons randomly assigned to the no-treatment comparison group were offered individual crisis intervention, a treatment factor that confounded study findings.
There is growing evidence to support the use of CBSM strategies to improve psychological well-being and physiological status in persons living with HIV disease. Findings related to therapeutic social support as a moderator of psychological stress also have been positive. Little comparative research has been done to determine the relative effect of these two types of interventions on either psychological or physiological status. This study was undertaken to compare the effects of CBSM groups, social support groups (SSG), and a wait-listed control group on the outcomes of psychosocial functioning (perceived stress, coping patterns, social support, uncertainty, psychological distress), quality of life, neuroendocrine mediation (salivary cortisol, DHEA levels), and somatic health (disease progression; HIV – specific health status; viral load; immune status [selected cellular cytokine production levels; NK cell function; CD3+/CD4+, CD3+/CD8+, CD8+/CD57+ (cytotoxic T lymphocytes); and CD3−/CD57+ (NK) cells]).
For the psychosocial variables, it was hypothesized that participants receiving either intervention, as compared with those receiving standard clinic care, would have lower levels of uncertainty, perceived stress, and illness-related psychological distress and would report a higher quality of life and more effective coping strategies. For the physiological variables, it was hypothesized that the intervention groups would have lower cortisol:DHEA ratios, improved immune status, lower viral loads, and a better HIV-specific health status.
A pretest–posttest repeated-measures design was used in this randomized clinical trial to examine the influences of CBSM group and SSG participation on a battery of PNI-based outcomes. Quota sampling was used to achieve appropriate sample representation by gender, at a ratio of 4 males:1 female (20%). Gender subgroups were next stratified by prebaseline CD4+ cell counts to equilibrate study groups by initial CD4+ counts and, indirectly, by stage of illness. Participants then were randomly assigned to (a) the 6-month wait-listed control group or (b) one of the two intervention groups and scheduled to begin the treatment with the next available group (with separate intervention groups for men and women scheduled to begin approximately every 2 months). For the two intervention groups, Time 1 measures were obtained immediately prior to the intervention; Time 2 measures were collected immediately on completion of the 8-week stress management interventions; and follow-up measures were obtained at 6 months postintervention for Time 3. For the wait-listed control group, measures were obtained at baseline (Time 1), 8 weeks later (Time 2), and at 6 months (Time 3), at which time participants entered one of the intervention groups by random assignment.
Participants were recruited from two mid-Atlantic U.S. metropolitan areas, with direct recruitment in a large, regional infectious-diseases clinic and indirect recruitment through newspaper advertisements and widely distributed study brochures. Eligible participants were at least 18 years of age, able to read and speak English, previously aware of their diagnoses of HIV disease, and deemed physically capable (through clinical screening for Karnofsky performance scores [Karnofsky & Burchenal, 1949] of at least 60) of attending the intervention sessions and completing a 6-month follow-up requirement. Potential participants were clinically screened and excluded for (a) significant psychiatric illness (i.e., psychoses, dissociative disorders, severe and/or unstable depressive disorders, and organic mental disorders) and (b) cognitive impairment, as evidenced by a score of less than 20 on the Cognitive Capacity Screening Examination (Jacobs, Bernhard, Delgado, & Strain, 1977). Enrolled participants could not be pregnant or taking steroids or immunomodulatory drugs (including cytokines, thymic derivatives, and antineoplastic agents, but excluding antiretroviral drugs). To control the initial effects of antiretroviral drugs on immunological function (especially CD4+ counts and viral load), enrolled participants were excluded from immunological and virological analyses for initiation of or significant changes in medication protocols. Prior to study enrollment, eligible participants were wait-listed until they had been stable on antiretroviral drugs for 1 month (cf. Kovacs et al., 1995; Rinaldo et al., 1991).
Enrolled in the study were 148 individuals, 29 females (20%) and 119 males, with an average age of 39.4 years (median =39.0). The distribution of the sample by race and gender reflected the target population, with approximately 55% African American and 43% Caucasian. Of the total sample, 33.8% (n =50) were asymptomatic (CDC category A), 35.8% (n =53) were symptomatic (category B), and 30.5% (n =45) were classified as having AIDS. The sample’s revised HIV Center Medical Staging System (rHCMSS) scores ranged from 0 to 39 (the full range of this HIV-specific symptomatology scale), with a mean of 18.3 (SD =12.1). Study attrition was within the expected range, with 112 participants completing the intervention groups or initial waiting period (76% retention) and 102 individuals completing the 6-month follow-up visit (69% retention). The attrition rate did not differ among study groups.
As indicated in the conceptual-empirical schema for the study (Fig. 1), five psychosocial instruments were used to measure aspects of the stress process, including the adaptational outcomes of psychosocial functioning and quality of life. The Mishel Uncertainty in Illness Scale (MUIS; Mishel, 1981, 1984) is a well-validated, consistently reliable measure of uncertainty associated with illness. Higher scores on this 33-item scale indicate higher uncertainty in the areas of symptomatology, diagnosis, treatment, relationship with caregivers, and prognosis (Mishel & Epstein, 1990). Alpha reliability in this study was .87.
Stress levels and coping patterns were assessed by the Dealing with Illness Scale (DIS; McCain & Gramling, 1992), as revised in 1994 (unpublished data). The DIS is comprised of a stress subscale, which enables respondents to indicate the desirability or undesirability as well as the personal impact of experienced events, and a coping sub-scale. The revised 40-item coping subscale was modeled on the Revised Ways of Coping Checklist (Vitaliano, Maiuro, Russo, & Becker, 1985). For these 40 of the original 60 items, psychometric analyses revealed three empirical factors related to maintaining a positive attitude, adapting to illness, and withholding, with items approximately equally distributed across the theoretical dimensions of problem-focused, emotion-focused, and appraisal-focused coping. Higher scores reflect more frequent use of the various coping strategies. Alpha reliabilities in this study were .84 for the stress subscale and .77–.86 for the coping subscales.
The revised Social Provisions Scale (SPS; Cutrona & Russell, 1987) was used as a measure of social support. The SPS is a 24-item, 4-point Likert-type measure of six social provisions or components of social support. In other populations the SPS has had excellent construct validity and internal consistency (Cutrona & Russell, 1987), but has had limited use with the HIV-infected population. The subscale score for social integration and the overall score for social support were used to examine the influences of the interventions, particularly the social support groups, on study outcomes. Alpha reliability for the current study was .90.
Psychological distress specific to living with HIV disease has been previously documented in the form of avoidant and/or intrusive thoughts related to the illness (Ironson et al., 1990; McCain & Cella, 1995; McCain, Zeller, Cella, Urbanski, & Novak, 1996; Perry et al., 1992) using the Impact of Event Scale (IES; Horowitz, Wilner, & Alvarez, 1979). Because of its specific nature and previous sensitivity to psychosocial interventions, the IES was used to measure the subjective impact of living with HIV disease. The IES is a 15-item instrument with response options that indicate how frequently within the past 7 days each distressing thought has occurred. Higher scores on the subscales of intrusive and avoidant thinking indicate higher psychological distress. Alpha reliability was .92 in this study.
One of the disease-specific extensions of the general version of the Functional Assessment of Cancer Therapy (FACT-G) scale (Cella et al., 1993), the Functional Assessment of HIV Infection scale (FAHI; Cella, McCain, Peterman, Mo, & Wolen, 1996) was used as a multidimensional measure of quality of life in people with HIV infection. The 55 items of the revised FAHI (version 3) are grouped into subscales of physical, social/family, emotional, and functional well-being; relationship with physician; and additional concerns specific to HIV infection. Higher scores indicate a greater quality of life. The FAHI has demonstrated excellent psychometric performance using the overall quality of life instrument (the FACT-G), including sensitivity to stage of illness and intervention (Cella et al., 1996; McCain et al., 1996). The HIV-specific subscale of the FAHI was revised to the version 3 format on the basis of those studies. Alpha reliability was .92 in this study.
Neuroendocrine mediation of the stress process was defined by salivary cortisol and DHEA levels. Salivary cortisol is a reliable indicator of the biologically active, unbound fraction of circulating cortisol. Salivary cortisol concentrations accurately reflect changes in blood levels within 2 min and are not influenced by salivary flow rates (Kirschbaum & Hellhammer, 1992). Unstimulated salivary cortisol measures at the time of peak circadian levels and prior to appreciable exogenous influences (i.e., immediately upon a.m. arising) have been shown to be consistent across samples (Kirschbaum et al., 1990), with normal early a.m. values of 0.13–1.0 μg/dL (Aardal & Holm, 1995).
Salivary DHEA has been validated as an indicator of the unbound fraction of plasma DHEA. Like salivary cortisol, unconjugated DHEA is liposoluble and rapidly diffuses from plasma to saliva (at a ratio of approximately 30:1), so that salivary concentration is not dependent on salivary flow rate. Normal early a.m. values for salivary DHEA have been reported in the range of approximately 1–11 ng/mL (Granger, Schwartz, Booth, Curran, & ZakariaL, 1999; Lac, Lac, & Robert, 1993).
Salivette® (Sarstedt) sampling kits were used by participants to collect saliva specimens immediately on arising on the morning of each data collection point. Arising times and sample collection generally were between 7:00 a.m. and 9:00 a.m., but a participant’s usual time of arising was deemed primary in order to match saliva samples with that individual’s normal circadian pattern. No participant in this study had other than the normal nighttime sleeping pattern. Cortisol and DHEA levels of cryopreserved saliva specimens were batch-measured for a given participant by radioimmunoassays (Diagnostic Products Corporation) in the core laboratory of a general clinical research center (NIH M01 RR00065).
In the context of an immunologically mediated disease such as HIV infection, symptomatology represents both an indirect measure of immune functioning and a clinically significant outcome indicator for stress-related impairment (Nott, Vedhara, & Spickett, 1995). HIV-specific health status was assessed using the revised HIV Center Medical Staging System (rHCMSS; McCain et al., 1998). The rHCMSS provides an ordinal scaling system for rating the severity of HIV-related signs and symptoms from 0 (asymptomatic) through 39 (AIDS, with severe complications). This scale addresses only physical illness and thus is not confounded by neurological, psychiatric, or immunological variables. The rHCMSS has demonstrated excellent interrater reliability (r =.90) and construct validity (McCain et al., 1998). The rHCMSS was scored on the basis of the participant’s history by advanced-practice nurses employed for this study and explicitly trained and monitored in its use.
Phenotypic subsets of CD3+/CD4+, CD3+/CD8+, CD8+/CD57+, and CD3−/CD57+ lymphocytes were measured to evaluate, respectively, intervention effects on helper, suppressor, and cytotoxic T lymphocytes, and NK cells. These cell types are critical in the pathophysiology of HIV disease, in that they initially maintain immune system functioning but incrementally lose their functional capacity as the disease progresses (Cohen, Kinter, & Fauci, 1997). Immunophenotyping was performed in a clinical immune monitoring laboratory, using tightly controlled techniques for two-color flow cytometry (with a Becton-Dickinson FACScan flow cytometer, FACScan research software, and monoclonal antibodies).
NK cell function may be of major prognostic significance in AIDS pathogenesis. NK cells spontaneously lyse virally infected cells but become functionally defective as HIV disease progresses, possibly as a function of diminishing IL-2 and/or IL-12 production (Brenner, Dascal, Margolese, & Wainberg, 1989; Chehimi et al., 1992; Sirianni, Tagliaferri, & Aiuti, 1990). Studies in our laboratory and others have documented that mononuclear leukocytes can be cryopreserved and reconstituted without significant loss of cellular activity. Thus, cell samples were cryopreserved so that matched samples from a given participant could be batch-processed, thereby reducing within-subject interassay variability (Fujiwara et al., 1986; Zeller, McCain, McCann, Swanson, & Colletti, 1996). After standard lymphocyte separation and cell-washing procedures, lymphocytes were resuspended in RPMI with fetal bovine serum. Following freezing to −80°C using a Forma model 700 controlled-rate freezer, cryovials of cells were stored in a liquid nitrogen freezer until they were thawed for the performance of assays.
NK cell cytotoxicity was evaluated in a standard 4-hr chromium (51Cr) release assay with NK-sensitive K562 target cells (American-type culture collection). 51Cr-labeled target cells were incubated together with effector cells for 4 hr in microtiter plates. Supernatants were then harvested, and the amount of radioactivity was determined by a gamma counter. NK assays are often performed using a range of effector cells (NK) to target cells (K562) (e.g., ratios from 40:1 to 5:1) to determine an optimal lysis ratio. In our laboratory, however, HIV-infected cells rarely exhibited K562 lysis above background levels at effector:target ratios less than 40:1. Thus, we used an effector:-target ratio of 40:1, and the results for NK cell cytotoxicity were compared as a percentage of lysis at this ratio.
To test the hypothesis that stress reduction enhances a type 1 cytokine pattern, levels of IFN-γ and IL-2 were measured to reflect a type 1 cytokine pattern and IL-4 and IL-10 to represent a type 2 pattern. The planned measurement of IL-12 was not possible because of frequently undetectable levels, and the measurement of IL-5 often required more participant cells than were available.
Leukocytes separated from peripheral whole blood samples were cryopreserved (as above) to enable batch testing of a given individual’s stimulated cytokine production levels. Lymphocytes recovered from cryopreservation were cultured in triplicate in a CO2 incubator. Cells were cultured in complete RPMI containing phytohemagglutinin (PHA) in accordance with the manufacturer’s protocol. Cytokine levels in the culture supernatant fluid were then measured using standardized enzyme-linked immunosorbent assay (ELISA) kits (Quantikine®, R & D Systems). IL-2 levels were measured after 24 hr of stimulation, and all other cytokines were measured at 72 hr. These incubation periods have been found to yield peak production levels of the respective cytokines in our laboratory and in others.
Host viral load is a critical factor in the immunopathogenesis of HIV disease, and peripheral viral load is consistently predictive of HIV disease progression and response to antiretroviral therapy (Fiscus et al., 1998; Hughes et al., 1997; Mellors et al., 1997; O’Brien et al., 1996; Vlahov et al., 1998; Voelker, 1996). With recent technological advancements, viral RNA has been sensitively measured in plasma by the quantitative-competitive polymerase chain reaction (QC-PCR) assay (or reverse transcription PCR). QC-PCR reflects viral replication as a function of the number of HIV RNA copies detected after target amplification. The Amplicor® HIV Monitor assay was used, with a lower detection limit of 200 RNA copies/mL of plasma. Assays were batch-processed in a clinical molecular diagnostics laboratory.
For both interventions groups of 6–10 individuals met for 90 min a week for 8 weeks. To facilitate group interaction, women’s groups were conducted separately from those for men. To remain in the study, participants were required to attend at least 6 of the 8 sessions.
Derived from Smith (1989; 1990) and Benson and Klipper (1990), the CBSM protocol consisted of skills training in focused breathing, progressive muscle relaxation, yogaform stretching, guided imagery, and beginning meditation, along with cognitive restructuring techniques and active coping skills. A fundamental assumption of the CBSM training approach was that different techniques are differentially effective for and preferred by different individuals. Hence, a variety of the most common skills were taught by personal instruction, followed by the expectation of at least daily practice of the skill for 1 week guided by professionally produced audiotapes. Once each skill was taught and given a “fair trial” of practice, individuals were assisted in personalizing relaxation approaches and combinations of techniques for ongoing stress management.
The SSG intervention, led by an experienced, doctorally prepared mental health nurse, focused on facilitating communication related to emotional issues, problem-solving and cognitive-reframing techniques, and individual and group empowerment. Specific goals (cf. Kelly et al., 1993; Leserman, Perkins, & Evans, 1992) included (a) developing and maintaining a positive attitude or “fighting spirit,” (b) cognitively re-framing stressful experiences to maximize personal growth, (c) planning a course of action to deal with illness experiences and other stressors, (d) expressing emotions concerning the experiences of illness, (e) sharing successful coping strategies, and (f) seeking and maintaining social support.
Repeated-measures analysis of covariance (ANCOVA) models were constructed using mixed linear model methodology (Brown & Prescott, 2000) to test the psychosocial and physiological sets of hypotheses that participants receiving either of the interventions would have, compared with those who received standard outpatient clinic care, (a) lower levels of uncertainty, perceived stress, and illness-related psychological distress and would report a higher quality of life and more effective coping strategies; and (b) lower cortisol:DHEA ratios, improved immune status, lower viral loads, and better HIV-specific health status. In addition to the design factors of group (CBSM, SSG, and WAIT), time (postintervention and 6-month follow-up), gender, and baseline CD4+ percentage, covariates included the baseline value of the outcome variable and baseline rHCMSS score. Thus, the critical intervention and prognostic variables used in the study design were included in all analyses.
The within-subject correlation from the repeated measures was modeled using a compound symmetric covariance structure. The homogeneity of covariance assumption, inherent to analysis of covariance, was tested by constructing the interaction terms of group by baseline score of a given outcome, group by baseline CD4+ percentage, and group by baseline rHCMSS. Interaction terms were included for group by gender and group by time to assess differential group means between the gender or time variables.
The order of the model building was as follows. First, the assumption of homogeneity of covariance slopes was tested for each of the three continuous covariates. When the tests of homogeneity of covariance slopes were associated with p values of less than .10, the heterogeneous slopes were modeled using a nested term. Next, the remaining interactions (group by gender and group by time) were examined, and interactions associated with p values of less than .10 were retained in the model. Finally, the baseline CD4+ percentage variable was dropped from the model if the p value was greater than .10. The variables for group, time, rHCMSS, subject, and outcome baseline were retained in the model because they were considered design variables. Group-level means were compared using a Tukey–Kramer multiple-comparison adjustment. When a model included heterogeneous covariance slopes, group-level comparisons were made at the minimum, mean, and maximum values of the covariate.
Because of the complexity of the mixed model and when the group or time effect was unclear, a univariate ANCOVA was used at each time point. The modeling strategy for the simple ANCOVAs was similar to the repeated-measures approach. Analyses were performed using PROC MIXED or PROC GLM in SAS version 6.12.
As analysis models were constructed, effects of design factors and potential cofactors were evaluated for each outcome variable. There were no significant differences found by gender (except for IL-4; see below). Means and standard deviations for major variables by group and time are shown in Tables 1 and and22.
Testing the psychosocial hypotheses using the mixed-model approach, significant group-level differences were found only with the FAHI and SPS. As seen in Table 3, significant group-level effects were identified in groups nested with the baseline rHCMSS scores. Interactions with health status were identified, which showed that SSG participants with higher rHCMSS scores had lower postintervention social/family functioning scores than did the CBSM and WAIT groups, as well as lower social integration scores than the CBSM group. In contrast, those participants with the lowest rHCMSS scores in both intervention groups had higher postintervention emotional well-being scores than did the control group.
With the physiological hypotheses (Table 4), there were significant findings only for NK cell cytotoxicity and cellular production levels of IL-4. SSG participants with higher rHCMSS scores had significantly lower NK cell cytotoxicity postintervention compared with the WAIT group and were marginally lower than the CBSM group. For the IL-4 variable there were group interactions with gender as well as rHCMSS, showing that (a) women had higher IL-4 levels postintervention than did men at all levels of rHCMSS, (b) at the mean level of rHCMSS, women in the SSG had higher IL-4 levels than did those in the CBSM group, and (c) men in the SSG group with rHCMSS levels at or below the mean had lower IL-4 levels than did those in the WAIT group. Because of the small subgroup sizes for women in this study, these findings must be viewed tentatively. There were no significant differences in cortisol levels among the groups. However, there was a marginally higher DHEA level in the CBSM group than in the SSG and WAIT groups.
Using univariate ANCOVA models to discern significant effects at particular time points, it was found that the CBSM intervention group had significantly higher postintervention emotional well-being and total quality-of-life scores than did either the SSG or WAIT groups, but this effect dissipated by the 6-month follow-up. At 6 months postintervention CBSM participants had lower illness-related uncertainty but higher intrusive and avoidant thinking patterns than did the SSG and control group participants. Among SSG participants, significant changes were lower social/family well-being scores immediately postintervention and lower social support scores after 6 months. There alsowere lower IL-4 levels among SSG participants in comparison with the CBSM and WAIT groups immediately postintervention, but not at 6 months.
When the effect of the symptomatology cofactor (rHCMSS) was considered, participants with less disease progression in both intervention groups had higher levels of emotional well-being—that is, those who were less symptomatic had enhanced perceptions of emotional well-being after 8 weeks in either intervention. However, when analyzed without regard to participants’symptom levels and at specific times, the CBSM groups were found to have a higher overall quality of life than either the social support or the wait-listed control groups immediately postintervention. At 6 months post-intervention, SSG participants perceived lower social support, CBSM participants had higher psychological stress in the form of distressing thought patterns, and the immediate postintervention enhancement in quality of life for the CBSM groups was no longer evident. Thus, with both interventions positive effects were not sustained over the follow-up period. These findings indicate the need for longer, perhaps more “potent” interventions and/or routine booster sessions following the short-term weekly interventions. CBSM participants had lower illness-related uncertainty at 6 months, suggesting an effect of intervention such as enhanced use of problem-focused or appraisal-focused coping strategies, although there were no statistically significant changes in the coping measure used in this study.
For those in the social support groups who had experienced more symptomatology, study findings indicated perhaps untoward effects of the intervention. Postintervention, these participants perceived lower social/family well-being as well as less social integration, indicating a negative effect of group participation in the domain of social functioning. It may be that among those with more advanced illness group interventions that involve discussing or confronting one’s illness are not helpful in managing chronic psychological stress in the current era of HAART. These data suggest that a comprehensive assessment of how treatment advances may have modified clients’ needs would be useful in determining what alternative approaches might be more effective as the nature of the disease and its treatment have changed.
Since the advent and widespread use of protease inhibitors, in 1996, the goal of HIV disease treatment has been to decrease morbidity and mortality by continuous maximum suppression of HIV replication, as indicated by maintenance of an undetectable viral load (Carpenter et al., 1998). In this sample increasing symptomatology, declining CD4+ cell percentage, and increasing viral load continued to distinguish disease progression in the early HAART era. Other than the expected correlation with CD4+ cells, there were no statistically significant findings related to viral load in this sample, in part because of successful viral suppression leading in our assays to viral RNA levels frequently being below the detection limit of 200 RNA copies/mL.
Lower levels of NK cell cytotoxic function with more advanced HIV disease would not be unexpected. Yet such a decline was seen at post-intervention only among social support group participants. Because CBSM participants did not have such a decline, it may be that the expected deterioration in NK cell functioning with advancing illness was attenuated in that group. These findings need further exploration and validation.
It has been suggested that psychological stress, by virtue of increased cortisol and decreased DHEA, may drive a switch from a type 1 to a type 2 cytokine pattern in HIV disease (Esterling et al., 1992). Thus, the marginally higher DHEA levels among CBSM participants might reflect a trend toward the inversion of the cortisol:DHEA ratio, such as was found by Cruess et al. (1999), that would be protective in HIV disease. Although there were no detectable differences in cortisol levels among the groups at any time point, this, as well as the DHEA levels, may be attributable to somewhat imprecise measures. Standardized ELISA kits now available (e.g., Salimetrics®) will yield more precise measures in future studies. In addition, cortisol and DHEA levels may need to be measured by averaging multiple samples to increase accuracy.
Although the lower IL-4 levels among SSG participants immediately postintervention are consistent with the hypothesis concerning reduction in type 2 cytokine levels, the pattern of cellular cytokine production levels in this sample did not support the general hypothesis concerning a switch from a type 1 to a type 2 cytokine production pattern with advancing HIV disease (cf. Clerici & Shearer, 1993, 1994). Previous studies by Clerici et al. (1992) demonstrated that HIV-infected individuals preferentially elaborate type 2 cytokines and at lower levels than do individuals with normal immunocompetence. These observations, along with previous behavioral intervention studies in persons with HIV disease, formed the basis for study hypotheses that stress management interventions would lower stress levels and improve quality of life, reduce cortisol levels, and improve immunocompetence by maintaining a type 1 cytokine profile or perhaps inducing a switch from a type 2 to a type 1 cytokine profile. However, analysis of immunocompetence and cytokine profiling in this study revealed a more complex pattern than anticipated, which may, in part, reflect the impact of third-generation antiretroviral therapies. Participants did not demonstrate a type 2 cytokine pattern, as has been historically observed among persons with HIV disease. Rather, a type 0 pattern was demonstrated, an indication of greater immunocompetence than expected. The selection of cytokines for this study was theoretically driven, but it may be that measuring other cytokines, particularly TNF-α and IL-6, as well as monitoring for a longer time, will reveal more information about cytokine patterns in the era of HAART.
Although the study was ostensibly underpowered, particularly for the Time 3 follow-up data, the clinically relevant differences we observed were, in fact, statistically significant. Comparison of the group means in Tables 1 and and22 with those in Tables 3 and and4,4, demonstrate that, for those variables with clinically meaningful levels of group differences (i.e., the FAHI social/family and emotional well–being subscales and the NK cell cytotoxicity levels), there also was statistical conclusion validity. Thus, the actual power of the study was adequate to detect clinically significant intervention effects. A second issue concerns the high levels of variance in the physiological variables. Although a certain range of analytic variability reflects normal physiological variability, the precision and accuracy of the assays themselves and the research controls to reduce variance remain critical issues for PNI-based research. Multivariate modeling and critical covariate controls are essential for clinically meaningful and statistically valid research in this complex paradigm.
These study findings highlight the need for ongoing evaluation of stress management interventions designed to address the specific needs of this population. Along with the findings of others, this work substantiates the need for inclusion of both psychosocial and physiological outcome measures with a particular emphasis on documenting new patterns of physiological response to behavioral and psychosocial interventions among persons with HIV disease.
A unique contribution of this study was the examination of psychophysiological responses in a cohort of HIV-seropositive individuals during the initial era of HAART. These pharmacological therapies have changed the HIV disease trajectory to one of a chronic rather than a fatal disease. As a result, the stressors for those living with HIV infection have changed. Importantly, individuals must now deal with complex, demanding medication regimens with significant side effects. In addition, those once facing death are now rebuilding their lives while coping with a chronic and still stigmatized illness. Interventions that can attenuate the stress response require further investigation because of their potential to affect not only the quality of life but also the illness trajectory. PNI provides a sound theoretical framework for investigating both the psychosocial and physiological complexity of living with HIV disease.
Contract grant sponsor: National Institute of Nursing Research, National Institutes of Health (Nancy L. McCain, principal investigator); Contract grant number: R01 NR04395.
Contract grant sponsor: General Clinical Research Center of Virginia Commonwealth University/VCU Health Systems (John N. Clore, projector director); Contract grant numbers: M01 RR00065, NIH.