A total of 456 participants were recruited to the trial from six corporations of different natures of business including real estate, insurance, property management, and other industries. Baseline data collected prior to the randomisation of participants in the intervention and control groups of the CRCT were complete without any refusals to respond to the survey. The descriptive statistics on the demographic, health, and work-related variables, and the outcome variables, namely work-related burnout were summarised in Table 1. As shown, the average age of participants was 38.3 years (s.e. = 2.4) with slightly more females than males (54.6% vs. 45.4%), with the majority attaining an education level of a university degree or above (n = 370), and less than half were married (48%). For work-related variables, all participants worked full-time except three, with an average duration in the industry of 8.3 years (s.e. = 1.1) and about one-third (n = 143) could work flexible hours. The majority of participants were non-smokers (95.6%), light or non-drinkers (92.2%), and more than two-thirds exercised regularly (n = 342), with only a small proportion (8.2%) requiring three or more days of sick leaves in the past three months. In terms of the work environment measures by the Woo’s Work Environmental Scale, the mean standardised scores of nearly all domains were close to 50 with 48.2 (s.e. = 2.8) for work involvement, 51.3 (s.e. = 1.6) for co-worker cohesion, 54.3 (s.e. = 4.7) and 50.1 (s.e. = 2.1) for supervisor support and work pressure, respectively. Workplace management control was the only domain that scored the highest with an average standardised score of 60.5 (s.e. = 5.0). However, slightly more than half (52.7%) of participants indicated that they had contemplated resigning from their company in the past three months. Regarding the outcome measures of this study, namely work-related burnout, the average scores of the three domains including emotional exhaustion, depersonalisation, and professional accomplishment were 21.6 (s.e. = 0.4), 6.9 (s.e. = 0.4), and 27.8 (s.e. = 1.1), respectively. These represented about 60% of participants rated at a moderate (29.2%) to a high level (31.5%) on emotional exhaustion, about 45% on depersonalisation (moderate/high: 29.9%/15.4%), and only 32% on professional accomplishment (moderate/high: 20.9%/11.0%) following the classification provided by the scale authors.

Results of the final models of the Multinomial logistic regression analyses on factors associated with the three domains of work-related burnout were summarised in Table 3. As shown, a slightly different set of variables was associated with different levels of the three burnout domains. For emotional exhaustion, three variables were retained in the final model, but were only significantly related to the high level of this domain of burnout. This included age, co-worker cohesion, and work pressure, with the former two negatively, and work pressure positively, associated with emotional xxhaustion (OR = 1.09, 95%C.I. = 1.05–1.13). These results suggested that an increase in age and co-worker cohesion was related to a decrease in the emotional exhaustion of workers, and an increase in work pressure was associated with an increase in emotional exhaustion. For depersonalisation, four variables were identified as being related to different levels of this domain, At the moderate level of depersonalisation, age, co-worker cohesion, and work pressure were significant in the same direction as with emotional exhaustion, having odds ratios of 0.95 (95%C.I. = 0.92–0.99), 0.98 (95%C.I. = 0.97–0.99), and 1.03 (95%C.I. = 1.00–1.06), respectively. At the high level of depersonalisation, a slightly different set of variables was observed, with age and co-worker cohesion remaining, and work involvement replacing work pressure. All three variables were negatively associated with the high level of depersonalisation with odds ratios of 0.88 (95%C.I. = 0.82–0.95), 0.97 (95%C.I. = 0.95–0.98), and 0.98 (95%C.I. = 0.97–0.99), respectively. In terms of the professional sccomplishment domain, two variables were retained in the final model, namely drinking and resignation intention. These variables were both positively related to the moderate level of the domain only with odds ratios of 2.87 (95%C.I. = 1.17–7.36) and 1.66 (95%C.I. = 1.10–2.51), respectively.

The bivariate relationships between the three domains of work-related burnout and participants’ demographics, health and work-related variables, and the work environment variables were investigated with the results summarised in Table 2. As shown, with only adjustment for the clustering effect, a different set of variables were associated with each of the three domains. For example, age, education level, and working duration were related to emotional exhaustion but not the other two domains. On the other hand, intention to resign and supervisor’s support was associated with all three. (Table 2) These variables, with others that satisfied the pre-set selection criteria, were included in further analyses.