Our primary analysis showed no significant difference in weight change between TAT and SS at 12 months post randomization. In addition, we observed no between group differences in changes in stress, depression, quality of life, and insomnia. Insofar as TAT, and other mind-body interventions, may be postulated to work through modulation of such factors, results suggest no difference in effect between TAT and SS for study participants.
Similarly, we found that while intervention session attendance predicted weight loss maintenance, self reported TAT home practice did not. The findings can potentially be interpreted as consistent with no effect of the TAT intervention on weight change, beyond the benefit of group support. On the other hand, practitioners of energy psychology techniques maintain that symptoms, illness, or counterproductive behavior may be caused by or attributable to suppressed trauma or emotions, and that, through proper application of the energy psychology practice, these factors can be healed, or definitively resolved. Through the lens of such a paradigm, more practice is not necessarily better. Further research is required toward clarifying the mechanism of action of TAT and other energy psychology techniques. Such an evidence based mechanistic framework will enable clinicians and investigators to better determine, prescribe, and measure the appropriate frequency of practice in future clinical trials.
In our study mean initial weight loss was 9.8 kg overall, with adjusted 12 month weight regain of 1.72 kg and 2.96 kg for TAT and SS respectively. These results are comparable with what has been reported in other long-term weight management trials. In the Weight Loss Maintenance Trial [21
], for example, mean initial weight loss was 8.5 kg, while 12 month weight regain for participants in the self directed (no treatment ) arm was 3.7 kg, compared with 2.69 kg weight regain for an interactive technology arm and 2.12 kg weight regain for the personal contact arm. For both TAT and SS participants, weight regain was thus relatively modest, and considerably less than the 9.8 kg of average weight loss initially achieved.
For the secondary analysis, the final regression model included three statistically significant predictors of weight loss maintenance: session attendance, education, and the interaction between treatment assignment and initial weight loss. The strongest predictor of weight loss maintenance was attendance at weight loss maintenance group sessions. The finding is not unexpected, given the strong association between session attendance and weight loss documented in our previous trials, [32
] and suggests the importance of the social support offered in groups, and potentially the facilitators. Similarly, the observed significant association between education and weight loss maintenance is not surprising, with better educated participants maintaining more weight loss at the 12 M visit.
The significant interaction term is of special importance. Because a significant interaction between treatment assignment and initial weight loss was found, a major assumption underlying the main effect test of treatment effect in the primary analysis model was not met, and the test on the main effect term is interpreted with caution. The significant interaction between treatment arm and amount of weight loss in the initial weight-loss program suggests a treatment effect that varied depending upon initial weight loss. Post hoc analysis was then indicated so as to identify what the TAT treatment effect was at various levels of initial weight loss. This exploratory post hoc
analysis suggested that participants who lost the most weight in the initial weight-loss program maintained more weight loss if assigned to TAT than if assigned to SS. As Figure shows, the estimated weight regain between 0-M and 12 month-follow up for SS increases with increasing weight loss in the initial weight-loss program, while the estimated weight regain between 0-M and 12-month follow up for TAT decreases with increasing weight loss in the initial weight-loss program. The direction of the association of these two variables in the SS arm is consistent with the relationship observed in data from the Weight Loss Maintenance Trial [21
]. In the TAT arm, however, this expected pattern was reversed, with greater initial weight loss predictive of improved
Explanations for this possible conditional treatment effect are speculative but may suggest direction for future research. Weight loss in the initial WLP may serve as a marker for other participant characteristics that we either did not measure or that we cannot measure as precisely as weight. It may be that those with greater initial weight loss were more motivated, and that TAT is most appropriate for highly motivated individuals. The TAT intervention does not have obvious face validity as a strategy for weight control, which may represent a barrier to its effective use, hence requiring a higher level of enthusiasm to overcome; however, we did not measure enthusiasm or motivation in this study. The TAT method may be a behavioral tool that is most effectively used by individuals who have undergone profound physical or psychological change [45
], such as substantial weight loss. Finally, it is possible that a longer follow- up time period may be needed to observe a TAT effect, with this effect first noticeable in those with higher initial weight loss.
In any case, caution is warranted in interpreting this exploratory finding, due to multiple potential confounders and other limitations. Differences in age, stress, depression, and other factors can potentially influence success in managing weight, and may have confounded results. Because participants are not blind to their treatment assignment, participant expectations can affect outcomes, yet we did not measure or analyze expectancy in our study. In addition, we were not sufficiently powered to account for nesting, or clustering, in our analyses, which may introduce additional bias. Likewise, our procedure for adding covariates to the augmented model toward arriving at the final model involves multiple tests, which could potentially introduce spurious findings.
The TAT intervention is a multimodal approach, and includes not only self acupressure, but also imagery, affirmations, drinking water, and other components. Although it was our intention to study TAT as a unified or whole system approach [46
], it is possible that just one, or several, of the individual components of the intervention could be responsible for any effect. Finally, although Table describes statistically significant model predicted differences between TAT and SS for participants at higher levels of initial weight loss, the effect sizes are small.
From a clinical standpoint, there as yet exists no real consensus on what constitutes a clinically significant reduction in weight. In terms of selected outcome measures for middle aged and older adults, 4 kg has been shown to be enough for a substantial drop in blood pressure (for borderline hypertensives) [48
], while a mean weight loss of 7% is enough for an effect on the probability of borderline diabetics developing frank diabetes [49
]. In addition, a mean weight loss of 5% has been shown sufficient for clinically significant improvement of mobility and quality of life in older arthritis patients [50
]. These data suggest that while the net weight change for participants in both study arms may be considered to be of potential clinical importance, any potential difference between the two arms is of uncertain clinical significance.