Results of the economic evaluations

Primary analysis
The calculated cost offsets associated with the provision of (one or two competing) psychotherapeutic interventions (PT) (i.e. Cost offset PT) and the additional cost-savings of implementing evidence-based psychotherapy compared to TAU (i.e. Cost offset PT vs. TAU) per patient per year, for each treatment approach separately, is shown in Table 1. Three studies [3,27,52] only provided data on costs either before or after the intervention; therefore no cost-savings comparison (i.e. cost- off set calculation) could be made. In addition, we did not attempt to provide the overall cost savings of TAU. Only two studies included in this review reported on pre- and post- cost data associated with the provision of TAU [25,26]. However, these findings are not sufficient to offset to any extent the cost-savings of the provision of TAU in the treatment of patients with BPD. Therefore TAU-related cost-data was not reported.

Cost offset PT
All studies comparing cost pre- and post intervention (15 evaluations; total sample size n = 787) reported a cost saving found to be in the range from USD $4 to $56,024 per person/per year. The pre-post analysis resulted in a mean weighted cost-saving of USD $5,840.92 [SD = $10,816.56; SE = $1,479.57] per patient per year (see Total Table 1 Cost Offset PT). Only the study by Palmer and colleagues (2006) [26] reported an increase of $99 in costs per patient/per year, post- intervention.
Because single group pre-post tests have recognised limitations, we have run additional analysis without the non-controlled studies to overcome bias of our results. Seven non- controlled studies were identified as so [28–31,34,36,54] with a total sample size of n = 192. For these non-controlled studies, the pre-post analysis resulted in a mean weighted cost-saving of USD $14,682.56 [SD = $ 1,286.07; SE = $17,820.38] per patient per year.
Subsequenly, re-running our primary pre-post analysis of the costs of psychotherapeutic interventions (cost offset PT) excluding the non-controlled studies yielded similar outcomes. Without the inclusion of these seven non-controlled studies the overall weighted cost offset of PT was: USD $2,987.82 [SD = $4390.31; SE = $180.01] per patient per year across these studies.

Cost offset PT vs. TAU
The 15 evaluations (total sample size n = 1,415) comparing the healthcare costs of psychotherapeutic interventions for BPD to TAU related costs, mostly reported cost-savings. See Table 1 Cost Offset PT vs. TAU. Compared to treatment as usual, the additional weighted mean cost-savings of implementing evidence-based psychotherapy was USD $1,551.37 per person per year [SD = $6,574.17; SE = $174.77]. These cost-savings ranged from $83 to $29,392 per person/per year. Five evaluations [41,42,45,47,50] did not find a cost-saving and reported an increase in costs when comparing healthcare costs of psychotherapeutic interventions for BPD to TAU related costs (increase in costs ranged from $61 to $10,772). See Table 1.

Secondary analysis
We furthermore were interested in whether there were detectable differences in the cost-benefit of different types of psychological therapy. We were able to find six studies of DBT and seven non-DBT alternative approaches. There are not enough studies to directly compare two alternative approaches to date, we therefore pooled the seven non-DBT studies. DBT was compared to TAU in six evaluations [33,40–43,45]. The (weighted) mean cost-offset derived from the provision of DBT compared to TAU was USD $78.43 [SD = $3,1412.83; SE = $184.01] per patient per year across these studies (total sample size n = 344). Seven non-DBT evaluations focused on the post- healthcare costs related to the provision of other psychological approaches, namely: MBT [25,47]; CI [48]; CP [49]; CBT-PD [51]; CBT-TAU [26] and MACT [50]. Compared to TAU, a (weighted) mean increase in costs of USD -$1,150.03 [SD = $5,482.33; SE = $170.91] per patient per year was demonstrated across the studies (total sample size n = 1029).
The weighted mean difference in cost savings vs. TAU derived after the provision of DBT vs. non-DBT studies was: $1,228.43 (SE = $251.13; 95% confidence interval $735.63 to $1,721.30). This resulted in a significant weigthed mean difference (t(953.33.) = 4.892, p = .000). Although the difference per patient is significant, it is important to recognise that treatments themselves begin from different cost bases–and DBT standard includes weekly 2.5 hours of group, one hour of individual and ancillary care and phone coaching–which is generally more intensive that the comparative treatments which are typically up to two hours individual per week. In addition, crisis care, because it diverts from hospitalisation, has a large cost-offset. Although there are many potential interpretations for this significant finding, it could suggest that there is a greater cost-benefit of non-DBT vs DBT approaches compared to TAU. However, caution should be used when interpreting these results.