Discussion
In young people who had recently repeatedly selfharmed, we found no clinical or cost benefits for family therapy over treatment as usual in terms of hospital attendance for subsequent repetition of self-harm. For interventions in which the whole family was the subject of the assessment, and more than one person is involved in treatment, methods to assess benefits (and harms) beyond the individual should be considered.30 Our finding that family therapy is cost-effective when considering combined benefits to the young person and caregiver is therefore salient, although it assumes QALYs can be aggregated across individuals as a simple sum. This addition has been done in previous studies of child health,34 but is not part of the NICE reference case and assumes interdependence between utilities (the health state) of the adolescent and caregiver. Our study sample had baseline levels of difficulty at least as severe as the average CAMHS referral.17 The proportion of female participants recruited and the mix of referrals from hospital and community sources are similar to those seen in other studies.4 The treatments given in the treatment as usual group were highly varied, which can be expected from a pragmatic trial, but broadly similar to CAMHS practice in the UK with a mixture of supportive or individual counselling, cognitive behavioural therapy, and family work;17 however, the proportion of patients who received formal family therapy for self-harm (87 [21%] of 417) was much higher than we were expecting. Those in the treatment as usual group who received a version of family therapy did not, however, have significantly better primary outcome results than either those in the family therapy group or those in treatment as usual who only received other interventions.