Demographic characteristics and clinical outcome variables
Recruitment and follow-ups took place over 12 months from November 2018 to November 2019. The demographic characteristics of the participants are summarized by group in Table 1. The mean age was 50.5 years (standard deviation = 10.9). The majority of the participants were female (70.6%) and had a secondary education level (55.1%). The descriptive statistics of the outcome variables at the three time points by group are also shown in Table 1. There were no significant differences in the demographic data and outcome variables at baseline when comparing groups. The growth trajectories of different outcome variables among three groups are depicted in Figs. 2 to 7.
Unconditional mean model
The results of the null model suggested that the development of a multilevel model is warranted because intercepts varied significantly across individuals (Wald Z = 8.778, p < .001), and the intraclass correlation suggested that about 77.3% of the total variability in DASS-21 lies between individuals. Thus, we could further develop multilevel models to explain the variability in intercepts within and between individuals.
Linear and quadratic growth model
The individual variations of the growth rates of different outcome measures in respective groups are presented in Table 2. The results of the parameter estimates of fixed effects of Group × time interaction are presented in Table 3.
Regarding the intervention effect of MBCT compared with WC, participants in the former group showed significant time effect and time × group interaction effects on DASS-21, SF12 (Mental), CPSS, and PSQI. In terms of the primary outcome, individuals in the MBCT group revealed significant decrease (β = -6.19, t = -4.03, p < .001) of mood disturbance and significant increase in the rate of change (β = 1.52, t = 2.12, p < .05). With regard to secondary outcomes, participants in the MBCT group indicated significant improvement of mental health status (β = 1.87, t = 3.62, p < .001), sleep quality (β = -1.27, t = -2.50, p < .05), and perceived stress level (β = - 3.32, t = -3.68, p < .001). Effect size statistics also indicated that the MBCT group had small to moderate effects (d = 0.25 to 0.56) on the above outcome variables at post-treatment and follow-up assessments compared with the WC group.
Compared with the WC group, participants in the HQCT group showed significant time and time × group interaction effects on DASS-21. Individuals in the HQCT group demonstrated significant decrease (β = -10.9, t = ‑7.45, p < .001) of mood disturbance and significant increase in the rate of change (β = 3.31, t = 4.96, p < .001). With regard to secondary outcomes, significant time × group interaction effects on both SF12 (Physical) and SF12 (Mental) were also noted. Effect size statistics also indicated that the HQCT group had small to moderate effects (d = 0.15 to 0.64) on the above outcome variables at post-treatment and follow-up assessments compared with the WC group.
Comparable intervention effects between MBCT and HQCT were noted across different outcome variables. In terms of primary outcome, both interventions led to significant reductions of mood symptoms, with relatively more reductions in the HQCT group during post-treatment (d = 0.24) and follow-up (d = 0.19) assessments. With regard to secondary outcomes, participants in the MBCT group showed much improvement in mental health status (d = 0.17 to 0.23) as compared with counterparts in the HQCT group. On the contrary, the latter showed much enhancement in physical health status (d = 0.12 to 0.40) across time.
Individual growth models with time-invariant and time-varying predictors
Table 4 shows the parameter estimates of the individual growth model of the MBCT group and HQCT group on DASS-21. Specifically, time-invariant predictors (gender, age, education) were nonsignificant in predicting linear or quadratic change of DASS-21 for both groups. For the MBCT group, SF12 (Mental) and CPSS were significant time-varying predictors of DASS-21. Particularly, participants in the MBCT group with increasing mental health status and decreasing perceived stress level tended to report an attenuation of mood symptoms over time. For the HQCT group, SF12 (Physical), CPSS, and GSE were significant time-varying predictors of DASS-21. More specifically, participants in the HQCT group with increasing physical health status, enhancing self-efficacy level, and lowering perceived stress level tended to report a reduction of mood symptoms over time.