Participants in both treatment arms indicated moderate psychological distress at baseline (M = 9.8, SD = 3.7), and levels of psychological distress were not significantly different between treatment groups at baseline (). The two treatment groups did not differ on any other measured characteristics at baseline (). As shown in , control participants had slightly higher CD4+ T lymphocytes at baseline, greater ARV use at post-intervention, and had a higher proportion of heterosexual participants, although these differences did not reach conventional levels of statistical significance (all p’s > .08). These differences would be expected to favor the control group and thus could not account for any positive intervention effects observed. It is also possible that these differences could reflect different study populations due to study dropout rates or other factors. To minimize any potential confounding effects, ancillary analyses were conducted (described below) to control for ARV use, number of days between the pre-and post-test assessments, sexual orientation, and age.
Baseline characteristics of the treatment sample participants
The average participant in the study was male, African American, homosexual, unemployed, and not on ARV medication. Participants on ARV medications indicated an average self-reported adherence of M = .45 (SD = 76) missed doses in the past week at baseline and M = .71 (SD = 1.14) at post-intervention. Participants who completed all assessments (N = 38) were compared to study dropouts (N = 29) on all baseline characteristics. Study dropouts were significantly younger than study completers, t(65) = 2.26, p = .03, but no other significant differences emerged. Dropout rates in the treatment sample in the 8-week MBSR condition (N = 8) and the 1-day control condition (N = 1) were not significantly different (χ2(1) = 2.10, p = .15). Although not significant, a greater proportion of participants in the 8-week condition dropped out, and is likely due to the more demanding and time intensive nature of this program.
To determine the effects of the MBSR and control programs, mixed effects linear models analyzed log-transformed and raw CD4+ T lymphocyte counts in both the treatment (N = 48) and randomized (N = 67) samples. A mixed effects linear model on the treatment sample revealed a significant time × treatment interaction on log10CD4+ T lymphocytes, F(1,45) = 5.70, p = .02 () and raw (non log-transformed) CD4+ T lymphocytes (), F(1,44) = 5.03, p = .03. Specifically, the 8-week MBSR participants had a raw mean increase of 20 CD4+ T lymphocytes, whereas the 1-day control participants had a raw mean decrease of 185 CD4+ T lymphocytes from baseline to post-intervention (). Analyses of the entire randomized (N = 67) sample, which included participants who never received any treatment, showed the same pattern of results, F(1,57) = 3.15, p = .08. The potential clinical impact of this finding was assessed by counting participants with CD4-defined AIDS (<200 cells/mm3) at post-intervention. No participants had CD4-defined AIDS at baseline; 14% (N = 2) of participants in the control condition reached CD4-defined AIDS at post-intervention compared to 0% (N = 0) of participants in the 8-week MBSR condition.
Log transformed CD4+ T lymphocytes at baseline and post-intervention in the treatment sample (N = 48). Vertical lines indicate standard errors of the means.
Baseline and post-intervention means and standard errors for CD4+ T lymphocytes in the 8-week and 1-day MBSR treatment participants (N = 48)
Ancillary analyses using mixed effects linear models explored the role of confounding factors on the time × treatment CD4+ T lymphocyte effect. The inclusion of ARV medication status at baseline and post-intervention (binary variables: on or off ARVs) as covariates did not affect the strength of the time × treatment interaction on CD4+ T lymphocytes (F(1,45) = 5.50, p = .02), indicating that mindfulness meditation training buffers CD4+ T lymphocyte declines in participants both on and off ARV treatment regimens. Additional analyses explored the potential confounding effects of variability in the number of days between the baseline CD4+ T lymphocyte measure and the initiation of the 8-week MBSR class, sexual orientation (0 = heterosexual, 1 = homosexual or bisexual), and participant age in three separate mixed effects linear models. Similar to ARV medication status, the time × treatment interaction on log10CD4+ T lymphocytes remained statistically significant after control for variability in the number of days between the baseline measure and the initiation of the 8-week MBSR class (F(1,45)=5.66, p = .02), sexual orientation (F(1,45)=5.64, p = .02), or participant age (F(1,45) = 5.51, p = .02).
Mediation analyses were consistent with the hypothesis that mindfulness meditation treatment adherence explained the observed time × treatment interaction on log10
CD4+ T lymphocytes. Treatment participants attended 12.5 (SD = 7.6) class hours in the 8-week MBSR program and 6 (SD = 0) class hours in the 1-day education control. As shown in , a series of mixed effects linear models assessed statistical mediation (Baron and Kenny, 1986
). In a model with the treatment sample (N
= 48) where class attendance and the time £ treatment interaction were entered simultaneously, MBSR class attendance predicted greater CD4+ T lymphocytes at follow-up (b = .014, t
(74) = 2.09, p
= .04), and the path between the time × treatment interaction and log CD4+ T lymphocytes was no longer significant (b
= .06, t
(56) = .56, p
= .58). This finding indicates that the frequency of class attendance could potentially account for as much as 2/3 of the total effect of the experimental conditions on CD4+ T lymphocytes.
Fig. 3 Mediation analyses in the treatment sample (N = 48). MBSR class attendance mediates the time × treatment interaction on log10 CD4+ T lymphocytes. Numbers represent b coefficients, with parentheses representing b coefficients when MBSR treatment (more ...)
As previous findings have found effects of behavioral stress management in reducing HIV RNA levels in participants with detectable levels of HIV RNA at baseline (Antoni et al., 2006
), exploratory analyses examined the effects of mindfulness meditation training on HIV RNA levels in the sub-sample of participants with detectable baseline levels of HIV RNA (N
= 30). No differences in baseline HIV RNA levels were observed between the 8-week MBSR (M
= 46,880 copies/ mL, SD = 12,830) and 1-day control sub-sample (M
= 12,700 copies/ mL, SD = 25,660) participants with detectable levels of HIV RNA (t
(28) = 1.19, p
= .24). Consistent with the limited statistical power available in this small sub-sample, no significant effects were observed on HIV RNA (time × treatment interaction, p