A routine outcome measure for youth mental health: Clinically interpreting MyLifeTracker

Abstract Aim MyLifeTracker is a session‐by‐session mental health outcome measure for young people aged 12 to 25 years. The aim of this study was to determine clinically significant change indexes for this measure that would identify developmentally appropriate thresholds. The study also aimed to determine expected change trajectories to enable clinicians to compare a client's progress against average rates of change. Methods Participants comprised young people aged 12 to 25 years from both a clinical and a community sample from Australia. The clinical sample was 63 840 young people that attended a headspace centre. The non‐clinical group was an Australian representative community sample of 4034 young people. Results Clinically significant change indexes were developed for MyLifeTracker specific for age and gender groups by comparing clinical and non‐clinical samples. Males and young people aged 12 to 14 years needed to reach higher scores to achieve clinically significant change compared to females and other age groups, respectively. MyLifeTracker expected change trajectories followed a cubic pattern for those with lower baseline scores of 0 to 50, whereas those with baseline scores of 51 and above had varying patterns. For those with lower baseline scores, expected change trajectories showed that stronger change was evident early in treatment, which then tapered off before accelerating again later in treatment. Conclusions The development of MyLifeTracker benchmarks allows the measure to be used for Feedback Informed Treatment by supporting treatment planning and decision‐making. This information can help clinicians to identify clients who are not on track or deteriorating and identify when clients are improving.

of a single factor structure, although the five items were also designed to be clinically useful individually. Overall, MLT measures the quality of life, with higher scores indicating higher levels of quality of life. It has been validated against measures of psychological distress, quality of life and wellbeing, and demonstrates appropriate reliability and sensitivity to change (Kwan, Rickwood, & Telford, 2018). MLT is currently implemented into an electronic data system used by the 110 headspace youth mental health services implemented across Australia . This electronic data system provides information to clinicians that are collected from clients prior to every visit and displayed to the clinician in the form of a graph over time of MLT scores. This reveals change over time that can be used by the clinician to ascertain treatment progress and can also be shown to clients during their session via a computer or tablet device.
MLT was developed to fill a measurement gap in youth mental health. Historically, outcome measures have been designed that reflect the traditional mental health service demarcation between the child and adolescent services, for those aged less than 18 years, and adult services, for those aged 18 years and above (Kwan & Rickwood, 2015). The growing implementation of youth mental health services internationally, which span the age range of 12 to 25 years, necessitates new measures (McGorry, Bates, & Birchwood, 2013).
MLT was designed to be suitable for session-by-session use by being very brief and comprising only five items. An essential next step for the development of MLT is the identification of age and gender norms, which improves the interpretability of such measures (Centofanti et al., 2018). This information is particularly important in youth mental health because adolescence and early adulthood are periods of rapid social, emotional and physical development when age and gender differences are pronounced (Donald, Rickwood, & Carey, 2014;Rickwood et al., 2015).
The development of session-by-session measures for routine outcome monitoring supports Feedback-Informed Treatment (FIT) approaches, whereby a clinician receives quantitative feedback on a client's progress to use in-session and help guide treatment planning (Bickman, 2008). FIT requires a measurement system that is easily completed by the client and allows instant feedback to the clinician (Hall et al., 2014). This provides clinicians with regular up-to-date snapshots of a client's mental health status and shows any changes since past sessions (Lutz, De Jong, & Rubel, 2015). Clinicians are then able to monitor if clients are progressing or deteriorating between sessions, and adjust treatment planning accordingly (Boswell, Kraus, Miller, & Lambert, 2015).
Such an approach can also allow clinicians to bring the measures into sessions and feedback progress to clients, which can be a powerful way to promote shared decision making (Reese, Norsworthy, & Rowlands, 2009). FIT has been shown to improve communication between client and clinician, increase the accuracy of diagnosis, enable quicker adjustments to treatment planning when required, provide stronger outcome effects and improve the efficiency of treatment (Bickman, Kelley, Breda, de Andrade, & Riemer, 2011;Carlier et al., 2012;Janse, De Jong, Van Dijk, Hutschemaekers, & Verbraak, 2017).
A valuable metric for clinicians to use in FIT is change scores, such as clinically significant change indexes, expected change trajectories and early warning signals. These can be calculated from session-bysession measures to provide evidence-based benchmarks for FIT and routine outcome monitoring systems. A clinically significant change is conceptualized as the process of a client starting treatment in the dysfunctional (clinical) population and leaving treatment no longer in this population (Jacobson, Follette, & Revenstorf, 1984). It is operationalized as a change in a client's outcome measure score showing that they are statistically more likely to be drawn from the functional distribution, having moved out of the dysfunctional distribution during treatment (Jacobson & Truax, 1991). When the dysfunctional and functional populations are identified, clinically significant change indexes can be calculated by finding the value where the two populations intersect. Reliable change can also be determined, which takes into account the reliability of the outcome measure, ensuring that change is not due to measurement error.
Change can be then categorized into four stages: Deterioration-when a client has reliably worsened; Unchanged-when no reliable change has occurred; Improvement-when a client has made a reliable positive change but still remains in the dysfunctional population and Recovered-when a client reliably improves and moves into the functional population (Jacobson & Truax, 1991).
A criticism of clinically significant change is that it can be an overly stringent measure of change, being based on diagnostic cutoffs. In naturalistic clinical settings, some clients may not be able to reach this threshold because they initially present in the functional population range (Wise, 2004). Other methods of monitoring change have been recommended; specifically, the use of growth curve modelling, which shows expected rates of change (Donald & Carey, 2017). This approach estimates a mean starting point (intercept) and average rates of change (slope) of the pooled sample trajectory; that is, withinperson expected change patterns (Singer & Willett, 2003). The method is particularly useful for exploring client change in naturalistic therapy settings as it can deal with data that are time-unstructured and unbalanced. This provides clinicians with an expected change trajectory, which can be compared with an individual client's trajectory to determine whether the client is within or outside expected rates of change, potentially indicating the cause for concern (Finch, Lambert, & Schaalje, 2001).
Research has increasingly focussed on detecting clients who are at risk of deterioration using early warning systems that are derived from expected change trajectories (Finch et al., 2001). An early warning is evident when a client's score drops below an identified threshold. It is recommended that these early warning signals be derived from the bottom-end percentage of the targeted population and the proportion of clients who reliably deteriorate in that population (Finch et al., 2001;Warren, Nelson, Mondragon, Baldwin, & Burlingame, 2010). An essential aspect of early warning signals is the ability to accurately predict clients who are responding poorly to treatment or are not on track(NOT) before therapy is terminated (Boswell et al., 2015). Some studies have evaluated the efficacy of these signals of deterioration, alerting clinicians to clients that are falling into the bottom 10% to 20%, demonstrating detection accuracy rates of 85% to 100% when used with adult clients (Lambert et al., 2002). Lower detection accuracy rates of 69% to 88% are seen when early warning signals are used with children and adolescents, which has been justified by the higher proportions of treatment failure when compared to adult clients (Cannon, Warren, Nelson, & Burlingame, 2010;Nelson, Warren, Gleave, & Burlingame, 2013;Warren, Nelson, & Burlingame, 2009).
Therapeutic deterioration is evident in up to 10% of adult clients (Lilienfeld, 2007;Murphy, Rashleigh, & Timulak, 2012), but much higher at 21%, for clients in youth psychotherapy settings (Warren et al., 2009). High dropout rates are another major concern in youth mental health settings, and dropout has been shown to be partly due to clinician and therapeutic factors that may be responsive to feedback (de Haan, Boon, de Jong, Hoeve, & Vermeiren, 2013). Early warning alerts have been shown to reduce deterioration from 20.1% in treatment as usual to 5.5% in feedback conditions for adult clients (Shimokawa, Lambert, & Smart, 2010). Feedback was also shown to double the proportion of clients with clinically significant improvement in NOT clients. Feedback to clinicians alone, and to both clinician and client, has been shown to significantly positively increase the rates of change in short-term adult NOT clients (De Jong et al., 2014).
FIT approaches are increasingly being advocated because clinicians have been shown to have low accuracy rates of predicting client deterioration during therapy when using their judgement alone (Hannan et al., 2005;Hatfield, McCullough, Frantz, & Krieger, 2010).
It is proposed that clinicians have a self-assessment bias which serves to maintain a positive self-image (Parker & Waller, 2015). For example, Walfish, McAlister, O'Donnell, and Lambert (2012) explored clinicians' ratings of their own clinical skills and client outcomes, showing that they rated their skills on average at the 80th percentile and that all clinicians rated themselves above the 50th percentile. In addition, clinicians on average believed that 77% of their clients improved as a result of their therapeutic intervention, which is well above the onethird proportion of clients shown to improve in most naturalistic settings (Walfish et al., 2012). Deliberate practice, incorporating FIT with evidence-based benchmarks, could be very effective at reducing this self-assessment bias amongst clinicians (Chow et al., 2015;Goodyear, Wampold, Tracey, & Lichtenberg, 2017;Macdonald & Mellor-Clark,-2015). Despite the potential clinical utility, however, clinicians have been shown to have limited knowledge around the use of routine outcome measures in predicting client deterioration (Bystedt, Rozental, Andersson, Boettcher, & Carlbring, 2014).
The current study investigated the implementation of routine outcome measurement and clinician feedback within youth mental health services, using the MLT measure. We aimed to determine MLT clinically significant change indexes that would identify developmentally appropriate thresholds for different age and gender groups. It was anticipated that there would be different clinically significant change indexes across the developmental period between 12 and 25 years and between males and females, due to the major changes that take place during adolescence and early adulthood and the marked gender differences in mental health status between males and females (eg, females displaying higher levels of psychological distress) (Brann, Lethbridge, & Mildred, 2018;Centofanti et al., 2018;Kwan et al., 2018). Identifying these developmental patterns would allow clinicians to provide more tailored client care. To do this, scores for a clinical population group were compared with data from a nationally representative community sample to determine appropriate change indexes. It was hypothesized that the non-clinical group would have higher MLT scores compared with the clinical group, that males would have higher MLT scores than females, and that the younger adolescents would have higher MLT scores than those who were older (Kwan et al., 2018). We also aimed to determine expected change tra-  .This sample consisted of 40.4% males and 59.6% females, in the following age ranges: 12 to 14 years (24.1%), 15 to 17 years (32.0%), 18 to 21 years (29.1%) and 22 to 25 years (14.8%).
The non-clinical group was a nationally representative community sample that consisted of 4034 young people aged 12 to 25 years from across Australia. The sampling was stratified to provide a near-even split between males (49.1%) and females (50.9%), and across age groups: 12 to 14 years (24.7%), 15 to 17 years (24.7%), 18 to 21 years (25.0%) and 22 to 25 years (25.6%).

| Procedure
The clinical group commenced their first episode of care at a headspace centre between July 1, 2015 and March 31, 2017. During this period, data were available for 101 headspace centres across Australia. headspace centres routinely collect a minimum dataset comprising data from young people and their service providers at every occasion of service. The dataset includes demographic characteristics, clinical presentation and treatment outcome measures. Young people can present for a wide range of reasons to headspace centres , but only those who were deemed by their clinician to be at one of the following stages of mental illness were included in the current analyses: mild to moderate general symptoms; sub-threshold diagnosis; threshold diagnosis; periods of remission or ongoing severe symptoms.
The data from headspace centres are encrypted and uploaded to a national datawarehouse, which is used for research, monitoring and evaluation. Ethics approval was obtained through quality assurance processes, comprising initial consideration and approval through the headspace board research sub-committee. The consent processes have been reviewed and endorsed by an independent body, the Australasian Human Research Ethics Consultancy Services.
The non-clinical group was recruited between July and September 2018. A research consultancy agency was commissioned by headspace to undertake a national computer-assisted telephone interview of young people aged 12 to 25 years from across Australia.
A quota sampling procedure was used to ensure equal numbers by gender and age group. The sample was recruited through random digit dialling (RDD; randomly generating Australian mobile phone and landline numbers). Ethics approval was obtained from Bellberry Limited Human Research Ethics Committee.

| Measures
Both the headspace minimum dataset and headspace nationally representative community survey include a large number of demographic, clinical and outcome measures. For the current study, only the demographic characteristics of gender (male, female, other), age group (12-14, 15-17, 18-21 and 22-25 years), and the MLT routine outcome monitoring measure were used.

| Routine outcome monitoring measure
MLT (Kwan et al., 2018) is a five-item self-report measure used to assess the current quality of life in areas of importance to young people. It asks young people to indicate how they have been feeling over the last week in relation to their: "general wellbeing (emotional, physical, spiritual)", "day-to-day activities (study, work, leisure, self-care)", "relationships with friends", "relationships with family" and "coping (dealing with life, using your strengths)". Responses are given on a sliding scale anchored at 0 and 100 with the chosen score visible, accompanied by a visual analogue of a sad and happy face as anchors. Total MLT scores were calculated by averaging across the five items, ranging from 0 to 100, with a higher score indicating a higher quality of life. In the present study, internal consistency was high, with the Cronbach's α = .83 in the clinical group and .88 in the non-clinical group. The original MLT study reported a Cronbach's α of .84, which ranged from .79 to .86 across age groups and gender (Kwan et al., 2018).

| Data analyses
SPSS V21 was used for all analyses. First, descriptive statistics for MLT were calculated and a factorial between groups analysis of variance (ANOVA) was conducted to evaluate the differences in MLT scores across population groups (clinical, non-clinical), gender (male, female) and age groups (12-14, 15-17, 18-21, 22-25 years). Games-Howell post-hoc tests were conducted to address unequal variances and sample sizes. Due to the large sample size, a significant change was reported as partial η 2 > .001 and d ≥ .02. Clinically significant change indexes were calculated using data from the clinical and non-clinical samples for each age group and gender (male and female; there were too few participants reporting nonbinary gender in the non-clinical sample to create a third gender group) combinations. Results from the original MLT study revealed differences in baseline MLT scores across age and gender groups (Kwan et al., 2018). The formula proposed by Jacobson and Truax (1991) was used to calculate clinically significant change indexes when both clinical and non-clinical groups are available but have unequal variances (p. 13).
Expected change trajectories were determined for the clinical group using growth curve modelling (Singer & Willett, 2003), which estimated average rates of change in MLT composite scores across participants' episodes of care. This approach was utilized as it provides fixed effects that estimate a mean slope of the pooled sample trajectory (withinperson patterns). Maximum likelihood estimation procedures were used.
Weeks in treatment were used over session number as the time variable because this has been recommended in the past literature exploring youth psychotherapy change  and provided a better model fit based on Bayesian Information Criterion (BIC).
Expected change trajectories were calculated for decile groups dependent on MLT baseline scores; that is, 0 to 10, 11 to 20, etc. A precedence has been set for this method by past research exploring change trajectories, which show differing rates of change dependent on baseline severity on outcome measures (Finch et al., 2001;Lambert et al., 2002). Only data from participants attending more than one session and with treatment length up to 26 weeks were used to avoid extreme outliers in terms of treatment length. Two early warning signals were calculated based on the baseline MLT score and expected change trajectory: one SD below the expected change trajectory and reliable deterioration based on the baseline MLT score. Table 1 provides the descriptives for MLT scores for the clinical and non-clinical groups, and the calculated clinically significant change indexes for MLT across age groups and gender. The ANOVA revealed no significant interactions (partial η 2 ≤ .001) and only significant main effects. MLT scores were significantly higher in the non-clinical group compared to the clinical group (partial η 2 = .149); and for males compared with females (partial η 2 = .005). MLT scores differed significantly by age group (partial η 2 = .013), and post-hoc analyses revealed that scores for those aged 12 to 14 years were significantly higher than all other age groups ( Table 2 shows the growth curve model slope estimates. The expected change trajectories followed a cubic pattern for those with a baseline score of 0 to 50; a quadratic pattern for baseline scores of 51 to 60; a linear pattern for baseline scores of 61 to 70; and non-significant change over time for baseline scores of 71 to 80. MLT baseline scores of 81 to 100 again followed a cubic pattern; however, this was inverse to change trajectories seen in MLT baseline scores of 0 to 50. Within baseline scores between 0 and 50, expected change trajectories for the lower scores showed a steeper increase (linear growth), greater deceleration (quadratic growth) and a bigger acceleration (cubic growth) compared with higher scores. A similar trend was evident for MLT scores between 81 and 100, but in the opposite direction, trending downwards.

| Early warning signals for use in clinical practice
Two early warning signals were calculated: the first was a growth curve one SD below the expected change trajectory (SD = 19.81, the yellow line in Figures 1 and 2), which would warn that the client had fallen below the 16th percentile of expected change while in treatment. The yellow line would be relevant only for MLT baseline scores of 0 to 70 as they have an increasing trend, and MLT scores for 71 to 100 would not be necessary as they would reach reliable deterioration before they dropped below one SD of the expected change trajectory.
The second early warning signal (red line in Figures 1 and 2) indicates when a client has reliably deteriorated from their baseline MLT score.
Reliable change has previously been calculated for MLT to be a change of 18.27 points, and reliable deterioration would mean the client has dropped 18.27 points below their baseline score (Kwan et al., 2018). The red line would be relevant for all baseline MLT scores.
T A B L E 1 Descriptive statistics for MyLifeTracker for the clinical and nonclinical groups, and clinically significant change indexes, by age group and gender   15 in the second session, but this score is still above the yellow line, which means it is within one SD of expected change. By session seven, her MLT score is above the expected change trajectory for young people with baseline MLT scores of 21 to 30. Her progress remains above the expected change trajectory, which indicates she is making similar or better progress compared with other young people in treatment who started with a similar MLT score. At sessions 10 and 11, the young person's MLT score is still under the clinically significant change index but her score has increased above the 18.27 points (reliable change) from her baseline indicating reliable "improvement". By session 13, she has an MLT score of 66, which is above the clinically significant change index, meaning that this young person has moved out of the clinical population. The change can be categorized as "recovered" as the young person has reliably improved and moved from the dysfunctional population into the functional range (Jacobson & Truax, 1991).
The second example, shown in Figure 3, shows a negative therapeutic change directory. It is of a 12 to 14 year old female with a baseline MLT score of 36.40. The clinically significant change index would be 68.33 and her expected change trajectory would follow that of clients with baseline MLT scores between 31 and 40. By session two, this young person has a score of 20, which alerts the clinician that she has dropped below the yellow line. In the third session, the young person has a score of 26, which brings her back above the yellow line, but by session four she dips back below the yellow line with a score of 21.40. In session five, the young person has an MLT score of 10.40, which indicates she has dropped below the red line and the young person remains below the red line for the remaining sessions.
In this example, the first early warning signal (yellow line) is triggered at two-time points, which tells the clinician that the client is dropping below one SD of expected change and that treatment planning may need to be reviewed. The second early warning signal (red line) is triggered by session five, showing the client has reliably deteriorated, and treatment planning and current support needed to be reviewed.

| DISCUSSION
The current paper aimed to develop a set of clinically significant change indexes, expected change trajectories and early warning signals to help clinicians to interpret MLT for young people aged 12 to 25 accessing youth mental health services. Using comparative scores from a nationally representative non-clinical sample, clinically significant change score benchmarks were able to be derived to assess client progress throughout treatment. Two examples were presented to demonstrate how the newly created benchmarks and early warning signals could be used to inform clinical practice. Table 3

Clinically significant change index
This index provides clinicians with information on whether a client is more likely to be in the non-clinical (above the index) or clinical population (below the index). It allows clinicians to see when a client moves from the dysfunctional to the functional population group during treatment, known as "clinically significant change". These indexes are calculated by finding the value where the non-clinical and clinical populations intersect. MyLifeTracker has clinically significant change indexes based on gender and age group (see Table 1). When a reliable change (18.27 points) is also considered, change can be categorized into four stages: • Recovered-when a client reliably improves and moves into the functional population • Improvement-when a client has made a reliable positive change but still remains in the dysfunctional population • Unchanged-when no reliable change has occurred • Deterioration-when a client has reliably worsened (see below in "Early warning signals-Red line" section) Note: If a client is above the clinically significant change index, a client cannot reach "recovered" and it may be difficult to achieve reliable "improvement" due to how high the client's score is and because they are already more likely to be in the functional population. The client can still show reliable "deterioration".

Expected change trajectory
This trajectory provides clinicians with estimates of average rates of change for clients. An individual client's trajectory can be compared with the average trajectory to determine whether the client is within or outside expected rates of change. These trajectories are calculated using growth curve modelling based on a clinical group during an episode of care.

Early warning signals-Yellow line
This yellow line provides clinicians with a warning when a client drops one SD below the expected change trajectory. This would mean that the client has fallen below the 16th percentile of expected change while in treatment and that treatment planning may need to be reviewed. These yellow lines are modelled on the same growth curve as the expected change trajectories, however, start from 19.81 points (one SD) below the client's MyLifeTracker baseline score (see examples in Figures 2 and 3). Note: The yellow line for MyLifeTracker is only relevant for baseline scores of 0-70 as they have an increasing trend, and scores for 71-100 are not necessary as they would reach reliable deterioration (red line) before they dropped below one SD of the expected change trajectory.

Early warning signals-Red line
This red line provides clinicians with a warning when a client has reliably deteriorated from their baseline score during their course of treatment. This may indicate that the client has increased risk or concerns, is not responding to treatment and may prematurely dropout from treatment. Clinicians should review treatment planning and check if additional supports are required. These red lines are calculated as18.27 points (reliable change) below the client's MyLifeTracker baseline score (see examples in Figures 2 and 3). Note: The red line would be relevant for all MyLifeTracker baseline scores. However, the red line will not exist when the MyLifeTracker baseline score is too low as a client's score cannot drop below 0 during treatment. support FIT implementation (Centofanti et al., 2018;Kodet, Reese, Duncan, & Bohanske, 2019;Mayworm, Kelly, Duong, & Lyon, 2020).
Young people are shown to have higher rates of deterioration and clinicians are shown to have lower rates of accurately predicting deterioration compared to adults in mental health treatment (Cannon et al., 2010;Warren et al., 2009 tice and provide feedback to clients, and also clinicians' own deliberate practice. Deliberate practice, which is a process of systematic effort to improve performance with the guidance of a supervisor, ongoing feedback relative to essential skills, and refinement and repetition of practice (Goodyear et al., 2017), has been shown to contribute to differences between clinicians in client outcomes, with the most effective clinicians engaging in 2.8 times more deliberate practice than other clinicians (Chow et al., 2015).
There are still mixed views among clinicians using FIT, however, and this seems to affect its effectiveness (Lucock et al., 2015;Lutz et al., 2015). De Jong, Van Sluis, Nugter, Heiser, and Spinhoven (2012) showed clinicians who used the measurement feedback provided to them had improved outcomes for those clients NOT. Specifically, female clinicians and clinicians reporting higher commitment to using FIT at the start of treatment were more likely to use the feedback provided from the measure. Further, clinicians who were more likely to trust feedback from sources external to their own opinion (low internal feedback propensity), had clients with faster rates of change compared to clinicians with a high internal feedback propensity. Clinicians with a strong focus on achieving success (promotion focussed) were more likely to achieve better outcomes using feedback when compared to clinicians who focus on preventing failures (prevention focussed) (De Jong & De Goede, 2015). At a service level, clinics that showed a better implementation of feedback systems were more likely to have measures completed and outcomes viewed by clinicians, which in turn led to a more positive impact on client outcomes (Bickman et al., 2016). Training is increasingly available in the area of FIT and future research should target how to improve clinicians' acceptability of feedback monitoring systems and how to enhance its implementation and effectiveness (Law & Wolpert, 2014).
The results of the current study should be interpreted in light of its limitations. Notably, the clinically significant change indexes, expected change trajectories and early warning signals were created for an early intervention mental health service for young people aged 12 to 25 years in Australia. Further research is needed to determine whether the benchmarks would apply to young people attending specialist or tertiary services. The indexes were developed using a community sample from Australia, and it is unknown whether similar MLT scores would be found in other countries. Replication in other regions of the world focusing on the development of youth mental health systems, like Canada, Ireland, the Netherlands and California, is warranted (McGorry, Trethowan, & Rickwood, 2019). Furthermore, the current study only explored expected change trajectories dependent on baseline MLT scores, as past studies have shown that this accounts for a significant amount of variance in the rate of change (Lambert et al., 2002). However, it may be important also to create expected change trajectories for other predictors, such as the client's diagnosis and presenting issues. For example, a study on substance abuse treatment found that while baseline mental health measures were a significant predictor of rates of change, employment status and baseline craving levels were also significant predictors of rates of change (Crits-Christoph et al., 2015).
In conclusion, the development of these MLT benchmarks is an important step to increase the clinical utility of the measure. MLT was originally developed to fill a gap in the availability of routine outcome measures for youth mental health services provided to adolescents and young adults. The availability of these benchmarks, including clinically significant change indexes and expected change trajectories, enhances the clinical utility and interpretability of the measure (Boswell et al., 2015;Donald & Carey, 2017). Providing benchmarks that are age group and gender specific is also critical for this age range when there is a substantial developmental change occurring in multiple domains. The clinical benefits of FIT are becoming more widely known and have become part of the agenda for the future progression of psychotherapy (Emmelkamp et al., 2014;Lutz et al., 2015). It is essential that such practices can be applied in youth mental health, where dropout and lack of clinical change are particularly problematic.
The implementation of routine outcome measures, like MLT, and the use of benchmarks that enable clinicians to determine developmentally appropriate change directories that reveal recovery, improvement, lack of change or deterioration, is essential to supplement clinical judgement to improve clinical practice and outcomes in youth mental health settings.