Stefan Elbers

53 Evaluating IMPT programmes over time Heterogeneity in Outcomes. The Cochrane's Q -tests for all outcomes at each of the three contrasts were significant, except for self-efficacy at post-follow-up. For all other contrasts, this indicates that the null hypotheses that these studies are evaluating the same effect was rejected (Higgins & Thompson, 2002). In addition, only except for self-efficacy at post- follow ( I 2 = 0%), all values were considerably high, with the majority of the values over 90%. These analyses support our decision to refrain from pooling the effect sizes. Rather, multiple different patient, study or interventions factors may account for this variability. The self-efficacy post-follow-up contrasts indicate a stable maintenance pattern across studies. DISCUSSION Summary of Findings Our first objective was to investigate the development over time of patients who participated in IMPT programmes. The results indicate that the majority of the patient cohorts significantly improved from pre to post-treatment. Importantly, this was mostly maintained at final follow-up, which is in contrast to typical triangular relapse patterns that have been observed in other health behaviour change efforts (Wood & Neal, 2016). Although the results indicate that pre-post effects of IMPT are generally maintained over time, the possibility of relapse for individual patients should not be neglected. Closer inspection of the distribution of individual cohort data, such as the post to follow-up physical functioning data of Silvemark et al. (2014) (SMD = 0.06, 95% CI = -0.15 to 0.28), reveals that 47% of the patients show a decrease in physical functioning, assuming normally distributed data and a pretest posttest correlation of r = .59. To increase the accuracy of these rudimentary estimates, publishing the datasets along with the study, would allow for more detailed analyses on patient relapse across studies. This is especially relevant when taking into account that IMPT programmes are often considered as treatment of last resort (Jeffery et al., 2011). Our second objective was to explore the study, patient, intervention and outcome heterogeneity of the included cohorts. In line with our expectations and with previous studies, we observed substantial methodological and statistical heterogeneity despite overlapping theoretical foundations, such as the biopsychosocial model (Geneen et al., 2017; Guzmán et al., 2001; Waterschoot et al., 2014). This heterogeneity can be explained by different policies, cultures, resources, and research traditions that have been influencing these treatment programmes over time (Kaiser et al., 2018; Kaiser et al., 2017). However, to our knowledge the current study is the first attempt to extract and categorize the individual treatment modalities of IMPT programmes to assess the treatment content heterogeneity in