The recruitment procedures developed for this pilot study appear to be a viable and robust means of inducting young people into a trial of UK secondary school-based counselling. In addition, attrition rates for randomised participants were acceptable; no major ethical or procedural obstacles emerged; and participants and professionals involved in the trial generally described their experience as rewarding. This suggests that the present protocol could be scaled-up to a fully-powered randomised controlled trial of counselling in schools. However, to reduce the numbers of participants excluded at assessment due to low mental distress scores, it would be advisable to assess only those who demonstrated relatively high levels of emotional distress at screening (for instance, an SDQ-ES score of 4 or more).
With respect to preliminary indications of the efficacy of school-based humanistic counselling, findings were mixed. On the one hand, change on the primary outcome measure indicated that the counselling was not efficacious in reducing levels of emotional distress; and average changes across all outcome measures indicated only a small effect. However, the intervention did bring about significant improvements in prosocial behaviour and there was a trend towards greater self-reported improvements.
One possible explanation for these findings is that humanistic counselling, in general, has a negligible overall effect, as some previous meta-analyses have suggested . However, the significant interaction between amount of improvement and level of distress suggests that this relatively small effect size may be related to the inclusion of participants in the trial with only moderate levels of initial distress. It is a well-established finding in the field of both child and adolescent  and adult [36, 37] mental health that more distressed clients demonstrate more change. Hence, although the sample size is very small, the present finding of a large overall effect size with young people meeting the cutpoint for MDE suggests that humanistic counselling may prove to have acceptable levels of efficacy if tested within a more severely distressed population. This suggests that, for future studies, it may be advisable to use a more stringent inclusion criterion for levels of mental distress, such as a score of 5 or greater on the SDQ-ES, or SDQ Total Difficulties within the abnormal range.
Given, however, that many of the young people who enter school-based counselling do not have such high levels of emotional or psychological distress , the present findings may suggest that such an intervention is not appropriate for this population. However, an alternative possibility is that it is helpful, but in ways that are not picked up by standardised measures of emotional and psychological distress. Support for such an interpretation comes from four findings in the present study. First, there was a trend for counselling participants to indicate significantly more improvements than those in the waiting list condition when problems were self-defined (SDQ-SR). Second, those with lower initial levels of psychological distress reported just as much satisfaction with the counselling as those with higher levels. Third, responses to the Adapted Change interviews (Lynass, Pykhtina, Cooper: A thematic analysis of young people's experience of counselling in five secondary schools across the UK, submitted) indicated that the most frequent changes following counselling were to do with greater feelings of wellbeing and improved relationships, rather than direct reductions in levels of psychological distress. Fourth, significant positive improvements in the counselling condition were found on the prosocial subscale of the SDQ. For future trials of humanistic counselling which involve non-clinical populations, then, it may be valuable to include more personalised measures of psychological change [such as the Goal Based Outcome measure, ], as well as measures that focus on positive mental wellbeing  and interpersonal relating.
Another possible explanation for the overall low effect size for counselling is the brevity of the period between assessment and endpoint. This was set at six weeks as an ethical safeguard for young people allocated to the waiting list condition, who may have found a longer period unacceptable. However, young people participating in the control arm of this trial did not report feeling disadvantaged by this allocation, and did not self-refer to the pre-existing school-based counselling service. In addition, around a quarter of the young people receiving counselling indicated that they did not feel they had completed their work within the six week limit. For these reasons, for future research, we would suggest that it is appropriate to extend the intervention period to a school term (10 to 12 weeks).
Finally, in attempting to understand the relatively low overall efficacy of counselling in the present trial, it is worth noting that participants in the waiting list condition appear to have fared relatively well, and considerably better than control participants in similar trials [e.g., [39–41]]. Evidence from the Adapted Change Interview with waiting list participants suggests two reasons for this. First, they tended to experience the assessment interview as a very helpful intervention in itself. Second, the promise of counselling in a relatively short period of time (six weeks) tended to instil in them a considerable degree of hope, expectation and motivation which, in itself, has been found to be of considerable benefit [40, 41]. Although such factors would be of relevance in any psychological therapies trial, the relative brevity of the current intervention may have made them proportionately more significant. Again, this would suggest that the present design would benefit from a longer period between baseline assessment and endpoint.
The low to modest alpha coefficient of the SDQ subscales in the present study, including the primary outcome indicator (SDQ-ES), is something of a concern. This may reflect the limited length of the 5-item SDQ subscales, and has been identified as a problem in other studies of the SDQ's psychometric properties [42, 43]. For future studies, therefore, measurement of the primary outcome may benefit from a longer measure to maximise reliability.
With respect to other limitations, the small sample size in this pilot means that all outcome findings must be treated with extreme caution. Confidence intervals are wide for all outcome indicators, and a non-equivalent distribution of participants across the two conditions is quite possible. The lack of formal procedures for rating adherence and assessing inter-rater reliability is also an important limitation, and means that the exact nature of the intervention being delivered cannot be verified. Findings from the Adapted Change Interview should be treated with particular caution given that the unstructured nature of the response format may have led participants to provide more socially desirable responses. A final limitation of the present study is the lack of extended follow-up.
Although, with respect to efficacy, the present findings are mixed, given the proliferation of school-based humanistic counselling services in the UK, we believe that it is essential to undertake a fully-powered RCT of this intervention. The procedures developed in the present trial are a viable means by which to conduct such a study. However, we would recommend the following modifications:
Adopt a higher inclusion criterion for level of mental distress;
Assess only those young people who, at screening, indicate relatively high levels of mental distress;
Extend the period from baseline to endpoint to a full school term (approximately 10 to 12 weeks);
Incorporate measures of wellbeing, interpersonal functioning, and a personalised measure of change;
Use a longer primary outcome measure to ensure inter-item reliability.