The session wants and needs outcome measure for adolescent Web-based therapeutic support: examining the use of personalised free-text responses

Despite the increasing prevalence of mental health difficulties in children and young people (CYP; NHS Digital, 2021), they often face tremendous challenges when it comes to accessing appropriate specialist support services. In the UK, the most recent epidemiological study of its kind showed that between 25% and 35% of CYP with a diagnosable mental health condition accessed specialist mental health support (Green et al., 2005). Although this research was conducted almost two decades ago, more recent studies suggest that ever-shrinking government spending budgets (Neufeld, Jones and Goodyer, 2017) lengthy waiting times (Wolpert et al., 2016) and rigid eligibility criteria for support (Department of Health, 2015; Wolpert et al., 2016) are barriers that continue to make access difficult. Thus, it is of little surprise that the demand for alternative sources of mental health support, and especially for online communication technologies, is rapidly increasing. Digital support platforms can ameliorate the geographical (Hanley, Prescott and Gomez, 2019), time (Malik and Coulson, 2011) and bureaucratic (King et al., 2006) obstacles so often associated with traditional therapy. Digital therapeutic support can be provided in numerous ways (Barak, Klein and Proudfoot, 2009), either synchronously, for example through live sessions with a professional, that mimic in-person services (De Ossorno Garcia et al., 2023), or asynchronously (e.g. through peer-support forums).

One therapeutic approach garnering research attention within the digital sphere is the use of single-session therapies (SSTs), also known as one-at-a-time approaches (Dryden, 2020). SST, in principle, is based on both the practitioner and the service-user treating the session as if it is going to be the only one, regardless of whether that is the truth. Each session should aim, therefore, to provide complete therapy within the single encounter (Cox and Campbell, 2003; Perkins, 2006). This type of approach can be particularly important with groups that might not have sustained contact with a professional. For instance, a study examining how young people might use mental health services in the UK found that the modal number of sessions was one, with 24% of cases being closed after this initial appointment (Wolpert, Hoffman, et al., 2012). Whilst SST has not been strongly advocated in the past, it has received a growing interest in recent years due to the acknowledgement of such trends (i.e. if people are using therapy services in this way, then practitioners need to develop skills to both offer and evaluate this kind of work).

Given the above, brief interventions, like SST, have been proposed as a way of broadening access for younger clients and reducing the detrimental impact that high drop-out rates have on the efficacy of therapy (Schleider and Weisz, 2017). The latter is an important consideration given that some groups of CYP, for example, boys, the very young, and those with low frequency problems (Edbrooke-Childs et al., 2021) may struggle to sustain engagement over multiple sessions. Despite this, the value of attending just a small number of sessions is demonstrably evident. Talmon (1990) found that whilst the modal number of attended therapy sessions was one, the one session, in isolation, was often beneficial. It is apparent, therefore, that a shorter yet comprehensive approach to therapy, such as SST, may be useful for those who, for whatever reason, value choice and flexibility (Perkins, 2006; Schleider et al., 2020). Maximising SST’s potential is thus vitally important.

What does SST tend to look like, and what types of concerns is it most useful for? SST can be offered either within in-person settings (e.g. Perkins, 2006) or through digital means (e.g. Bambling et al., 2008). For either delivery method, service-users can schedule an appointment for the session ahead of time, or make use of a “drop-in” system. Positive outcomes have been linked with both (Hymmen, Stalker and Cait, 2013). SST tends to take a solution-focussed approach from the outset (Perkins, 2006). The practitioner’s task is to quickly establish “what the client wants to walk away with at the end of the session” (Young and Dryden, 2019, p. 645), and subsequently align the session as closely as possible to this. This strategy differs from typical therapeutic approaches, which, at outset, focus on assessing the service-user and their symptoms. True problem-solving only features considerably later (Perkins, 2006). SST has shown promising potential for reducing symptoms of depression and general distress (Hymmen, Stalker and Cait, 2013), plus anxiety and conduct problems (Schleider and Weisz, 2017); however, it may not be appropriate for those who present with psychotic illnesses or suicide ideation (Hymmen, Stalker and Cait, 2013). Research suggests that the benefits of SST for CYP can be maintained over time. In one study, the short-term benefits reported after one month were still present after 18 months had elapsed (Perkins and Scarlett, 2008). Further, high client satisfaction with SST has been reported by both CYP (Perkins, 2006) and adult (Harper-Jaques and Foucault, 2014) clients.

Despite the tangible benefits of SST provided above, a suitable and consistent method of documenting their therapeutic outcomes is missing from the current evidence base (Hymmen, Stalker and Cait, 2013). Nomothetic, standardised measures commonly used to harness the outcomes of therapeutic support have several qualities that render them unsuitable for use with very short interventions. By their nature, they cannot sufficiently capture progress resulting from SST: oftentimes, they are designed for use with multiple-session therapy, where more than one perspective (i.e. practitioner, patient and parent) is typically taken (Wolpert, Ford, et al., 2012). As such, in SST, measurement before and after the intervention has typically not been possible, with services having to rely upon measures of satisfaction instead of a validated measure of whether individuals’ goals have been met. In addition, many frequently used symptom-specific outcome measures cannot capture the immediacy of change that is so crucial when only one session is involved, and when there is no guaranteed follow-up appointment (De Ossorno Garcia et al., 2021). These issues with demonstrating the effects of SST may present an even deeper problem when sessions are delivered digitally. Digital SST is often anonymous and/or of a “drop-in” nature (De Ossorno Garcia et al., 2021). Users are therefore completely unknown to the practitioner, with little pre-existing information available with which to guide the one-time session. These factors further complicate the meaningful measurement of therapeutic outcome, especially when nomothetic approaches to doing this are applied.

The choice to attend SST often stems from a specific therapeutic aim, or a desire to resolve an immediate, tangible problem (Dryden, 2020; Hanley et al., 2020; De Ossorno Garcia et al., 2021). It therefore follows, as Young and Dryden (2019) stated, that the immediate wants and needs of the service-user should be the primary drivers of these sessions. Considering this, and the previously discussed widespread unsuitability of commonly used therapeutic outcome measures, de Ossorno Garcia et al. (2021) recently developed the Session Wants and Needs Outcome Measure (SWAN-OM): an instrument designed for use in SST for CYP. To briefly summarise the mechanisms of usage, CYP engaging with SWAN-OM firstly express up to three session wants and needs. These can be chosen from a pre-set list of items, or alternatively, a “free-text” option is available for users to enter their own personalised therapeutic motivation for attending the session. The practitioner views these immediately before the SST begins, an element that helps the service user to immediately steer the work that follows. The initial wants and needs are then revisited post-session, and the user records how successfully they felt that the session met each of their expressed wants or needs. The interface of the SWAN-OM, as shown on a mobile device, can be seen in Figure 1.

The personalised free-text want or need articulation method was added to the SWAN-OM following qualitative feedback from CYP during the measure’s development and initial user testing phase (De Ossorno Garcia et al., 2021). Its inclusion provides an idiographic element to the measure. In contrast with standardised, nomothetic outcome measures, idiographic measures are closely tailored towards the service-user and to their unique reasons for seeking this kind of support. Detailed nuances relating to individual experiences may be missed when only standardised, nomothetic measures are used to capture them (Beltz et al., 2016; Ashworth, Guerra and Kordowicz, 2019). This presents a problem for research as well as practice, as service-user voice tends to be lost or diluted when standardised measure completions are aggregated for review or evaluation (Jacob et al., 2021). Idiographic measures have also demonstrated greater success than nomothetic ones when capturing therapeutic change for CYP (Edbrooke-Childs et al., 2015), and furthermore, can be important tools in the shared decision-making elements of therapy (Flannery and Jacob, 2020).

Despite the apparent value of idiographic outcome measures, and the shortfalls of nomothetic approaches, which were outlined earlier, this does not mean that the latter should be disregarded entirely. A systematic review of idiographic progress measures by Lloyd, Duncan and Cooper (2019) reported that nomothetic outcome measures still play an important role, especially when conducting wider service or treatment-level evaluations, whereas idiographic goal measures tend to be a more useful way of monitoring therapeutic progress and outcomes. Further, contemporary research also reflects the strong psychometric properties of goal-based outcome measures (Duncan, Cooper and Saxon, 2023). There are however a number of issues that present challenges for measuring outcomes idiographically, including difficulties with drawing meaningful comparisons between service-users, and indeed over time for the same user, if the user is not choosing from an established set of constructs (Sales et al., 2021). Personal preference may also play a role for service-users and patients in terms of which method they feel the most comfortable articulating through. Some may find it challenging to identify their own wants and needs (Sales et al., 2021), and as a result, might appreciate the prompts given by pre-defined choices. Some may simply be less keen to self-articulate, preferring the ease and reduced cognitive demand afforded by selecting from a list (Alves, Sales and Ashworth, 2016). The division between standardised and idiographic tools is often viewed as a mutually exclusive dichotomy. However, researchers are beginning to advocate for a hybridised (Sales and Alves, 2012) approach, combining the strengths of both (Ashworth, Guerra and Kordowicz, 2019). A measure like SWAN-OM, where users have a choice over how they articulate their wants or needs and guide their therapeutic progress, may be one such example of this.

Good concurrent validity between the SWAN-OM and other commonly used and standardised outcome and experience measures (e.g. experience of service questionnaire (ESQ, Brown et al., 2014) and the positive and negative affect schedule (PANAS, Watson et al., 1988), has recently been identified (De Ossorno Garcia et al., 2023). These measures were chosen for comparison based on their relevance, item similarity and similar level of immediacy following exposure to a service or experience and several significant correlations were found when comparing at a total and at an individual item level. Despite these promising findings, it remains pertinent, therefore, that offering young service-users a flexibility of choice in how they articulate their therapeutic needs is demonstrably beneficial. Offering a personalised option like this should not be detrimental to therapeutic outcome. Thus, we must determine whether wants and needs articulated through idiographic free text differ from nomothetic pre-set options in terms of the likelihood of meeting them during SST. It may also be important to identify whether there is a particular type of service-user who is more inclined to choose free text to express a therapeutic want or need. The latter could be considered in terms of age, gender or ethnicity, providing useful insights into who benefits the most from the ability to personalise their session wants and needs.

As mentioned earlier, up to three wants or needs can be expressed per SWAN-OM completion. These can also be completed during each meeting, with up to nine sessions being possible. To date, it is unclear whether articulating multiple aims, both within one SWAN-OM and across multiple completions, influences the likelihood of resolution. Whilst not directly pertaining to free-text personalisation, this is an imperative gap to fill in establishing whether sessions that include an idiographic element are treated by users asSST rather than as a multiple-session therapeutic course.

In terms of how and why the SWAN-OM is used, a recent report by the Child Outcomes Research Consortium (CORC) found that the most commonly expressed wants and needs related to a desire to learn coping strategies, explore one’s emotions, or share one’s story (Ruby et al., 2021) during support. However, it has not been established whether motivations behind support-seeking differ when they are expressed through a personalised response rather than when selected from pre-set options. To bolster this understanding, a bottom-up content analysis of the topics articulated through free text will enlighten us as to how the function is interacted with. This evidence base is needed to ascertain the usage patterns of the free-text feature, and indeed to justify this idiographic element’s continued inclusion in the SWAN-OM.

To fill these identified knowledge gaps, the present study used data produced by routine usage of the SWAN-OM by 10–25-year-olds. We took a mixed-methods approach to answering the following key questions:

1. Did pre-set and personalised free-text wants and needs differ, in terms of the extent to which they were recorded as met during SST […]:

   • Within sessions where a free-text want or need was chosen alongside a pre-set want or need?

   • Between sessions where only pre-set wants or needs were selected, compared with those where at least one want or need was articulated through free-text?

2. Do any demographic characteristics differentiate service-users who chose to articulate their want or need through free text?

3. Does the articulation of multiple wants and needs influence their likelihood of resolution in SST?

   • Did the number of articulated wants and needs per SWAN-OM influence how likely they were to be met in SST?

   • Did wants and needs articulated by users who completed multiple SWAN-OMs differ from those articulated by users who only completed one SWAN-OM in terms of how likely they were to be met in SST?

4. How were wants and needs articulated through SWAN-OM’s free-text option? An exploratory, data-driven approach was taken to:

   • Qualitatively analysing the wants and needs articulated through free text, in an inductive manner; and

   • From this exploration, quantitatively analysing any differences in SST outcomes between qualitatively identified types or groupings of free-text response.

A cross-sectional research design was used to analyse a data extraction from routine SWAN-OM usage. A mixed-methods approach was taken towards investigating the lines of enquiry outlined in the research questions. However, the weighting of each research strand within the overall strategy gave the present study the status of “quantitative-dominant” (Johnson, Onwuegbuzie and Turner, 2007) research. After quantitatively comparing the therapeutic outcomes of free-text and pre-set selections (i.e. whether the service-user reported the want or need as met), qualitative content analysis was incorporated to explore the content of free-text wants and needs. The resultant topics and observations were then used to guide further statistical analyses (Creswell and Plano Clark, 2018).

SWAN-OM was developed to establish whether the wants and needs of CYP who make use of digital mental health support through Kooth are met. Kooth is a Web-based counselling service for UK-based service users aged between 10 and 25. It is free of charge at the point of delivery and requires neither a referral nor a waiting list, to access. Central to the service’s ethos is the humanistic ideal that “users are the central decision-makers in their journey towards well-being” (Noble et al., 2021; p. 9). Such a perspective reflects a sentiment evident within a long history of client-centred perspectives of counselling and support (Rogers, 1951). Further, it also reflects the growing body of research reflecting the importance of young people having agency and choice within therapeutic relationships (Cirasola and Midgley, 2023). This tenet extends to the service’s provision of support that focusses predominantly upon goals, wants and needs, as opposed to taking a clinical or diagnostic therapeutic approach (De Ossorno Garcia et al., 2021). The service recently published a Theory of Change (Hanley et al., 2020, 2021). Within the exploratory work leading towards the development of this framework, it was highlighted that existing measures of change were not capturing outcomes in line with Kooth’s humanistic philosophical underpinnings, these measures commonly focusing upon symptom reduction. To respond to this, it was therefore suggested that additional idiographic measures were used alongside these existing measures to complement this data and support therapists in their work. Whilst commonplace goal-based outcome measures were integrated into more sustained therapeutic work, as indicated above, an appropriate tool for use in SST was not identified. This recognition led to the development of the SWAN-OM, a measure developed for use within digital therapeutic support that can be accessed sporadically.

This project involved the examination of routinely collected anonymous secondary data. As this involved information that was shared by young people, ethical review was required by the academic institution of the first and second author of this paper. The project received ethical approval from the University of Manchester’s Research Ethics Committee in May 2024.

The SWAN-OM is administered over the course of an SST “chat” session. Prior to chatting with a practitioner, the user is prompted to select between one and three statements that reflect what they hope to achieve in the session. Then, 20 pre-defined statements (e.g. “talk about something I haven’t told anyone before”), split over five overarching themes (e.g. “share my story with someone”), are available to choose from, plus a free-text option can be used if they do not feel that the statements adequately cover their aim. Post-session, the user then rates the extent to which each of the chosen aims, including free-text choices, were met, on a Likert scale from −2 (strongly disagree) to 2 (strongly agree). This rating is referred to in the results section of this paper as “SWAN-OM score”. Within the Kooth service, the SWAN-OM can be completed a maximum of nine times per user. After this, a review of the contact is undertaken and, alternative, longer or more structured, forms of support within the platform are offered if deemed appropriate. Demographic information is collected when a user registers with the service.

The data extraction, spanning January to August 2021, comprised 1,255 SWAN-OM completions by 1,009 users. The start of this period represented the first availability of the free-text function and this continued until the end of August. All users who agreed who completed the SWAN-OM and agreed that their responses could be used for research purposes were included in the analysis. The extraction included the age, gender and ethnicity of the user completing each SWAN-OM, plus how many and which wants and needs were selected. If the free-text option was chosen by the user, their response was available for our qualitative analysis, and the corresponding SWAN-OM score for each want or need was included for comparison. From these scores, we calculated average scores for each user, both overall and for pre-set items only.

About 32.4% (n = 407) of SWAN-OMs within our sample were completed by CYP aged between 11 and 14, 54.8% (n = 688) were aged between 15 and 18 inclusive and 12.9% (n = 160) were aged between 19 and 25 inclusive. Further, 77.5% (n = 973) SWAN-OMs were completed by CYP identifying as female, 14.7% (n = 185) by males and 7.7% (n = 97) by those identifying as gender non-binary (agender or gender fluid). In terms of ethnicity, 80.7% (n = 1,012) of SWAN-OMs were completed by a CYP who listed their ethnicity as white, plus 6.5% (n = 81) Asian, 4.8% (n = 59) mixed, 4.5% (n = 57) other ethnicity and 3.6% (n = 48) black.

Please note that the operationalisation of gender in this study refers to user-selected gender identities, rather than to biological sex (Lindqvist, Sendén and Renström, 2021). Hence, Kooth does not treat gender as a dichotomous entity within their interface. Ethnicity is also a user-selected characteristic and is therefore also treated as an identity.

Initial exploratory analyses were conducted to establish whether the data fulfilled parametric assumptions. The observed variables violated the assumptions of homogeneity of variance, and random sampling (the sample was not random as all users gave permission for their data to be used for research purposes). Thus, a series of non-parametric tests were used to investigate the first three research questions. Following these, qualitative analysis was conducted on the free-text responses. The statements lacked the length, depth and experiential content warranted for a more thorough and procedural form of analysis, such as reflexive thematic analysis (Braun and Clarke, 2021). Therefore, a data-driven content analysis was undertaken instead. Firstly, the first author read each statement in turn, to establish the nature of the content within. This was noted down for each and treated as a code. These codes were then summarised into broader topics (e.g. “mental health”, “LGBTQ+”). As these topics were developed, previous ones were revisited and amended iteratively, to make sure that the data was categorised in the best possible way. Secondly, it was noticed during analysis that the wants and needs appeared to vary based on the articulation of immediate risk. Owing to this, each statement was assigned a risk level for an additional level of analysis – these levels were: high risk, unclear risk and low risk. The status of “high risk” was assigned to responses that discussed suicide, self-harm or abuse, or if the user expressed desperation or urgency. A third and final apparent distinction within the data was that some free-text wants and needs were extremely specific and clearly articulated, whilst others were comparatively vague. This was deemed a dichotomy worthy of exploration. How clearly a free-text response was articulated might influence the success of the SST. If the practitioner is presented with an obvious aim, they may be better able to direct the conversation accordingly. Articulating a clear want or need might also imply that the user is more certain of what they hope to achieve in the session. This, again, could make attainment more likely. This follows the reasoning of the stages of change (Prochaska, DiClemente and Norcross, 1992), whereby an individual moves from basic awareness that a problem exists, to eventually being able to identify and articulate the exact problem. It is only once this latter stage is reached that changes can be made to remedy it. In a counselling context, an individual at this stage is likely to benefit the most from goal or problem-focussed therapy. Considering this, the free-text responses were divided again into whether a clear therapeutic aim for the SST was articulated or whether the response was of a more exploratory nature, perhaps more suited to a longer course of therapy. All qualitative analyses were sense-checked by the entire research team (EB, TH, SDOG, LS), with consensus drawn on how best to categorise the data logically and meaningfully. As mentioned earlier, free-text comments and responses often lack the length, depth and coherent flow of other forms of qualitative data such as interview data. However, free-text responses should not be dismissed in terms of their suitability for qualitative analyses, and their diversity and richness can indeed add value to research (Rich, Chojenta and Loxton, 2013). These distinctions (articulated risk and clarity of wants and needs expressed in free text) were further examined with non-parametric tests, to establish if these factors related to SWAN-OM score, how many wants and needs were expressed per chat/SWAN-OM and the total number of SST chats the user had.

We found that 1,185 SWAN-OMs were completed using only pre-set wants and needs, and 70 used the free-text function either without or alongside pre-set selections. Below we detail our investigative findings. The numbering used in this section corresponds to that of the research questions listed at the end of the introduction.

In SWAN-OMs where a free-text want or need was articulated alongside at least one pre-set want or need (n = 46), a Wilcoxon signed-rank test found no significant difference between SWAN-OM scores given to either type of articulation, z = −1.375, p = 0.169. Post hoc power analysis found that there was a 90% chance of detecting any statistically significant differences with this sample size. When comparing those SST chats where only pre-set wants and needs were chosen (n = 1,185) with those where at least one want or need was expressed through free text (n = 70), a Mann–Whitney U test found no significant difference in SWAN-OM scores, z = –0.767, p = 0.443. There was a 98% chance of detecting a significant effect based on the sample sizes of this test. It is worth reiterating here that users can select up to three wants or needs per SWAN-OM, with a SWAN-OM score given to each. This means that SWAN-OM scores were averaged where appropriate for these two analyses. For example, of the SWAN-OMs where only pre-set wants and needs were selected (n = 1,185), the score used in the analysis was an averaged score if a user had given more than one want or need. Of note is the fact that the data used to explore this research question could be considered to violate the “independence of observations” assumption associated with these tests since some individuals completed it multiple times. However, the nature of SWAN-OM as a single-session therapeutic method means that each completion is treated as separate rather than dependent on previous or subsequent completions. For this reason, we deemed it inappropriate to consider the data to be nested.

Chi-square tests of independence were used to establish whether the characteristics (age group, gender and ethnicity) of CYP who expressed at least one want or need through free text differed from those who chose only pre-set options. To prepare the data, age was grouped into the following categories: 14 and under, 15–18 inclusive and 19–25 inclusive. These represent the varying life stages and unique challenges a person faces at each time, namely, secondary school transition and puberty, summative examinations and early adulthood. The likelihood of responding with a free-text response was significantly related to age, X² (2, n = 1,255) = 7.90, p =0.019. Post hoc analyses, using a Bonferroni-adjusted alpha value for multiple comparisons, found that SWAN-OMs completed by those aged 14 and under were more likely than those completed by other age groups to contain a want or need articulated through free text (p = 0.007), and simultaneously less likely to contain wants or needs articulated through pre-set items only. The distribution of articulation type across gender, X² (2, n = 1,255) = 0.45, p = 0.800, and ethnicity, X² (2, n = 1,255) = 5.25, p = 0.263, did not significantly vary.

Three wants and needs were selected in 64.5% (n = 809) of the 1,255 SWAN-OM completions, with 19.8% (n = 248) choosing to articulate two, and 15.8% (n = 198) choosing just one. A Kruskal–Wallis H test found that SWAN-OM scores (with each analysed score derived from the average SWAN-OM score of all wants and needs expressed in each chat) significantly varied depending upon how many wants and needs were expressed, H (2) = 7.973, p = 0.019. Post hoc Mann–Whitney U tests (Table 1) were used to identify the nature of the difference, by comparing the SWAN-OM score difference between 1 and 2 expressed wants or needs (Comparison 1), followed by 1 and 3 (Comparison 2) and 2 and 3 (Comparison 3). However, the necessary use of a Bonferroni-adjusted alpha value of p = 0.016 meant that no differences were considered significant.

As mentioned earlier, users can complete up to nine SWAN-OMs, in keeping with the ethos of SST. This is another way in which multiple wants and needs can be expressed. About 68.4% (n = 859) of SWAN-OMs were completed by a one-time SWAN-OM user, and 31.6% (n = 396) were completed by users who returned to use the measure more than once. Table 2 shows the distribution breakdown of multiple SWAN-OM completion, which shows that most users completed the measure only once.

A Mann–Whitney U test found no significant score* difference between SWAN-OMs completed by one-time, or multiple SWAN-OM users, z = 1.525, p = 0.127. A Spearman’s rho correlation was used to further explore this research question, to establish whether there was a relationship between the exact number of SWAN-OMs completed per user (Table 2) and the scores* of the SWAN-OMs they completed. A very weak significant correlation was found, r = 0.062, p = 0.028: this weakness might imply that the significant result was amplified by the large sample size.

* Each analysed score in these instances were derived from the average SWAN-OM score of all wants and needs expressed in each chat.

Please note that all extracts given within this section are composite examples. They reflect the essence of real responses given by users, but due to ethical considerations, they have been reworded and as such are not genuine participant quotes.

Content analysis of the 70 free-text responses revealed that 18 topics were covered within. Some responses (n = 19) were assigned more than one topic as per their content. As is clear from Table 3, free-text wants and needs relating to mental health and coping were articulated the most readily. With regard to mental health, general guidance was sometimes sought, for example “I want some advice about my mental health”. However, specific disorders were also mentioned: “I would like to talk about my anxiety”.

Eleven responses were categorised as “unclear” in terms of topic. Such articulations were usually of an open and exploratory nature, often expressing a desire to talk or explore options despite not mentioning a specific topic. Wants and needs such as “I just want to talk about things” or “I want help working out what to do next” were expressed.

As discussed in the data analysis section of this paper, where the rationales behind the distinctions made below are explained in further depth, the data was also categorised into risk level. About 18.6% (n = 13) free-text wants or needs were labelled as high risk, which included responses such as “I am struggling to control my anger”, or “I want to learn how to cope with my self-harm urges”. An “unclear” risk level was assigned to 30% (n = 21) of responses. Insufficient information was provided to ascertain the user’s current risk, in wants and needs such as “I need to talk about my mum” and “I want to discuss my bad habits”. Without knowing what concerns the user has about their mother, or what their bad habits are, their level of immediate risk cannot be decided. Lastly, a status of low risk was assigned (51.4%, n = 36) where a response clearly articulated a non-dangerous situation, like “I want to feel less anxious when I talk to the girl I like”, or “I feel unmotivated in school and I want to learn how to overcome this”. A Kruskal–Wallis H test was used to examine whether risk level influenced the extent to which that want or need was achieved within the SST. No significant difference in SWAN-OM score distribution across assigned categories of risk was found, H (2) = 0.715, p = 0.699. Further, Kruskal–Wallis H tests found that risk level was not significantly associated with the number of wants or needs expressed within one SWAN-OM, H (2) = .2.134, p = 0.344, nor was it associated with the number of separate SWAN-OMs completed by the user, H (2) = 2.884, p = 0.237.

The final distinction in the data warranting deeper analysis was the specificity/clarity of the want or need that was articulated with free text. Responses were divided accordingly into clear, target-focussed aims, and those that were exploratory in nature, and/or less clearly defined. About 47.1% (n = 33) of free-text responses along the lines of “I want to find out about the ADHD diagnosis process”, or “I want to learn about how to stay safe online” were deemed to contain a clear therapeutic aim. The remaining 52.9% (n = 37) of responses were classified as less clear or exploratory. For example, “I want to talk through how I feel with someone who understands”, or “I just want to feel better”. Similar to the different levels of risk identified, this distinction was analysed quantitatively. Three Mann–Whitney U tests were conducted, finding no significant differences between articulation focus and SWAN-OM score, z = 1.627, p = 0.104, the number of wants and needs expressed within one SWAN-OM, z = –0.527, p = 0.598, or the number of separate SWAN-OMs completed by the user, z = 0.334, p = 0.738.

The nature of single-session therapeutic support presents several unique challenges when it comes to measuring its value. This is particularly true when measuring clients’ progress towards actualisation of their motivations for seeking such support. The present study examined data from SST delivered to young service-users in a Web-based format. It explored the usage and outcomes of engagement with an idiographic free-text method of therapeutic aim articulation, which is positioned alongside a nomothetic equivalent within the same instrument. Several valuable observations arose from our analyses, which will now be explored.

RQ1, RQ2 and RQ4 involved close examination of free-text responses, including comparison with pre-set items. No significant differences between pre-set and free-text responses existed in terms of SWAN-OM score – the outcome that was treated as proxy for whether a want or need was met in SST. This was true when the difference was explored within SWAN-OMs (where an individual had made use of both articulation methods) and between SWAN-OMs (those where only pre-set items were given were compared to those where at least one want or need was expressed through free text). These findings tentatively suggest that wants and needs articulated through free text are equally as likely to be met in SST as those expressed through opting for a pre-set response. Thus, the idea that giving users a choice in how they choose to provide information in therapy is worthwhile (Alves, Sales and Ashworth, 2016), and that uniting idiographic and nomothetic measures to create a “hybrid” tool is both feasible and useful (Sales and Alves, 2012), is supported. In our research questions, we did not provide a one-tailed prediction of what we expected to find here. This was because we were unsure as to how exactly this idiographic element would be interacted with in SWAN-OM, despite its addition having been suggested by potential service-users (De Ossorno Garcia et al., 2021). Free text might have allowed the expression of more precise needs that were perhaps more detailed than those available in pre-set. More detailed expression could have led, in turn, to better actualisation of these wants and needs. On the other hand, unrealistic or vague session wants and needs might have been typed, perhaps restricting their actualisation potential. This distinction echoes the stages of change (Prochaska, DiClemente and Norcross, 1992), whereby the extent to which a service-user’s approach to therapy is target-focussed or exploratory might influence the outcome of this therapy.

Although no difference was found between free-text and pre-set responses in terms of therapeutic outcome, meaning that we cannot attribute Prochaska et al.’s (1992) theory to the different articulation methods, this topic of want and need precision was raised once more during the qualitative analysis of the free-text responses. Around half of free-text responses were defined as clear and problem-focussed, with the other half categorised as less clear or exploratory. This suggests that free text is chosen for several reasons. A user might feel that pre-text options are too broad to accurately express what they wish to discuss, or alternatively, they might be unsure what they want to talk about. Typing a broad topic of their choosing might, therefore, appeal more. The quantitative analysis of these different interactions showed that no one way of using free text resulted in a better outcome than the other. This finding suggested that both ways of articulating a want or need through the idiographic element of SWAN-OM were as useful as each other, and that the measure appears to cater equally well to a variety of articulation abilities and styles. Additionally, response clarity did not differ depending on the expression of numerous wants and needs, whether within one SWAN-OM or through the completion of multiple. Whilst this suggests that expressing several wants and needs does not influence how clearly free-text responses are articulated, further qualitative investigation is warranted to look at how pre-set and free-text responses relate to one another, particularly within the same SWAN-OM. For example, it would be pertinent to explore whether free text is commonly used to elaborate upon or provide additional direction to a pre-set selection, or conversely, if an entirely unrelated want or need, extraneous to their pre-set choices, is what is typically expressed. Looking at articulation content when a free-text response is that users only expressed session want or need may also prove insightful. These investigations, which were beyond the scope of the present study, would provide vital context to free-text articulation and even deeper insight into how this idiographic element is interacted with.

In terms of topic, the most frequent wants and needs expressed by free text related to mental health and coping. Users often sought advice on how to handle their emotions and reactions and wished to discuss their mental health to varying degrees of specificity (De Ossorno Garcia et al., 2023). Despite the prevalence of these topics, the level of immediate risk expressed in free-text responses varied, with 18.6% categorised as high risk. The fact that the remaining 81.4% expressed either a low or unclear risk level could tentatively suggest that CYP do not tend to use Web-based SST to deal with extreme distress, or when they are in an emergency situation. However, it must be acknowledged that risk status cannot be fully ascertained from the content of a free-text response alone, and there is nothing to suggest that those at higher risk cannot articulate through pre-set items. Therefore, these findings should be treated simply as risk articulated within an item, rather than as a clear division between users who are or are not at risk. Baring this in mind, our quantitative examinations of the risk level assigned to free-text responses, and whether this related to SWAN-OM score or the expression of more wants and needs, showed no differences. The former finding therefore implies that responses with high articulated risk are equally as likely to be met as those categorised as non-risky. Nonetheless, it clearly remains vital that clinicians who practice this type of therapy are fully equipped to detect and deal with harm and risk prevention, in those who do choose to approach SST with these needs. Comparison with pre-set needs is again needed to further this understanding of why free text is chosen, particularly the extent to which topics and urgency covered in these responses are extraneous to those that can be expressed by pre-set.

Demographics were also explored, to detect whether certain characteristics made a service-user more likely to articulate a want or need through free text. Neither gender nor ethnicity influenced articulation method choice. However, we found that younger users, aged 14 and under, were more likely to choose free text and less likely to choose only pre-set responses. This suggests that the ability to choose one’s own SST aim was especially valued by this age group, one which represents a crucial and tumultuous period in a young person’s life (Patalay and Fitzsimons, 2018). Fourteen is also the age by which approximately half of mental health conditions have their first onset (Jones, 2013). There is, however, reason for optimism for these youngest SWAN-OM users. Those who recognise the value of support-seeking in their early teens tend to show higher functional independence in early adulthood (Szwedo et al., 2017). Although we saw that idiographic methods of articulating wants and needs were particularly valued by younger users (they chose this option the most readily), alternative explanations must be considered. It may be the case that the pre-set items are either difficult for younger users to comprehend in terms of their wording, or that they simply do not relate to them as strongly. Continuing to ensure that language is reading age appropriate, and that pre-set aims cover the variety of age-related issues faced from between the ages of 10 and 25, may warrant further consideration. But nonetheless, keeping the free-text element in the measure so that wants and needs can be expressed in the user’s own words will continue to ensure relevance and appeal to all (Sales and Alves, 2012; Sales et al., 2021). The present study showed, after all, that it was an equally valuable method of want and need articulation.

RQ3 examined the entire data set, which included wants and needs articulated through either free text, pre-set items or a mixture of both. We found that users who expressed either one, two or three wants and needs did not differ in terms of how useful they found the session. Although we must therefore conclude that approaching an SST session with several different wants or needs does not impact whether these needs are met, Table 1 shows a general trend towards lower SWAN-OM scores where more wants and needs are expressed. Owing to the use of an adjusted alpha value, this trend was not significant. However, had this been possible, the additional power offered by parametric tests might have confirmed this trend.

The comparison of single versus multiple SWAN-OM completers, which again spanned the entire data set, also yielded a non-significant finding. Grouping data into those who completed one SWAN-OM and those who completed two or more was a suitable decision, despite a user’s ability to complete the measure up to nine times. This is because users completed an average of 1.24 (SD: 0.77) SWAN-OMs, supporting what can be seen in Table 2, that most completed it either once or twice. Owing to the drop-in nature of the sessions, users might return multiple times for one of two reasons. They might return because their needs were not met in previous sessions, or alternatively, because they might wish to discuss another separate issue or concern. These two approach motivations, despite not being recorded by the measure, are likely to result in quite different SWAN-OM scores: those with unmet needs that are discussed several times might consequently rate their satisfaction as lower. Conversely, those who approach on separate occasions with a new issue to be discussed in each might not rate the sessions dissimilarly to those who completed the measure only once. This means that the differences in SWAN-OM scores between (a) users who completes nine SWAN-OMs and users who completes one, and (b) users who completed two compared to one, might not be stark. Owing to the modal completion rate being one, followed by two as the next most common, the latter, (b), is more reminiscent of the actual comparison made within this study. The Spearman’s correlation went some way towards supporting the idea that multiple SWAN-OM completions did not influence scoring. However, as very few users completed a high number of SWAN-OMs, it may be the case that neither test appropriately explored this research question. Again, had parametric tests been possible, inputting completion rate into a model as a covariate may have yielded a more conclusive finding. Despite the questions raised by this investigation, particularly in terms of re-approach motivation, we found that most users did complete the measure only once. We can therefore see that the single-session nature of this type of support has been sufficiently captured in the data. The SWAN-OM, as a hybrid idiographic and nomothetic instrument, was, overwhelmingly, used sporadically by the sample, as intended. This usage pattern provides an element of validity needed to place this study firmly within the body of literature pertaining to brief interventions of this kind.

Although our research findings have been critically examined thus far within this discussion, there are several general limitations to the present study that should be considered. As was already raised, the non-parametric nature of the statistical analyses prevented the use of more complex exploratory models, as well as potentially limiting the effects that could be detected. Whilst such circumstances are difficult to remedy, this warrants deeper investigation into the questions raised within this discussion, and perhaps fine-tuning the variables included in such tests for maximal robustness. Such fine-tuning also applies when considering our qualitative analysis. The thought processes and categorisation methods were sense-checked by the entire research team in line with best practice for studies of this kind (Graneheim and Lundman, 2004), and the potential value of free-text responses (Rich, Chojenta and Loxton, 2013) should not be undermined. However, owing to the small sample size, the variability of the richness we found in the free-text data may have limited our ability to meaningfully categorise them in terms of topic or risk. In future, when more free-text data is obtained through the on-going use of the measure, conducting another qualitative analysis might result in more robust and diverse categorisation. In spite of this, the groupings remained our “best guesses”, and after all, when subjectivity is appropriately acknowledged, it should be treated as a core tenet of qualitative research rather than a true limitation (Galdas, 2017).

The measure was designed for use with CYP, in an anonymous, Web-based, pluralistic therapeutic setting. Owing to the increasing popularity of therapy delivered in this format, the findings may be tentatively considered by those wishing to integrate idiographic measures into similar services. However, it is important to acknowledge that given the SWAN-OM’s target demographic, further research and measure development must be conducted before any transferability to adult service-users can be assumed. A similar picture is true in terms of potential application to other forms of therapeutic service delivery, including in-person.

When examining demographic information, we found that ethnicity and gender did not influence the likelihood of opting to articulate through free text. However, we found that 80.7% (n = 1,012) of total SWAN-OMs were completed by a user who identified as white, and 77.5% (n = 973) were completed by users identifying as female. Owing to the nature of the chi-square test, this imbalance does not impact the reliability of the measure used. However, we acknowledge that had our sample been more representative, with a larger range of free-text responses, our findings may well have differed: this lack of diversity undoubtably influences the generalisability of this study to those groups that were underrepresented. It is pertinent that services, including the one from which the present study’s data was obtained, should take steps to address any evident inequalities in their provision. These inequalities in support access include not only age, gender and ethnicity but also include socioeconomic (Ahmad et al., 2021) and disability (Camm-Crosbie et al., 2019) statuses to provide just two more examples. When addressing these inequalities, services should be mindful of how any number of these characteristics can intersect with one another to further compound barriers preventing access to appropriate support (Fagrell Trygg, Gustafsson and Månsdotter, 2019).

This study provides insight into the usage of SWAN-OM, an SST outcome measure, with particular attention given to an idiographic want or need articulation facility. The results show how multiple use of the SWAN-OM, or multiple want and need selection, has no effect on score variability. Whilst we found no indication that free-text articulations were more likely to be met than pre-set choices, we suggest that they might be equally as useful, especially for younger users who tend to choose this option more readily. Potential explanations for these findings are postulated. As suggested above, deeper qualitative investigation into the nature of free-text responses, and how they both relate to, and are combined with, pre-set responses, is needed. The sample size of free-text responses in this study was low (n = 70), so any suggestions made in this study should be considered in light of this. Further investigation into the use and validation of hybrid idiographic and nomothetic outcome measures is also recommended across the wider therapeutic field. Overall, our findings provide useful emphasis on the value of personalisation items in digital instruments and Web-based therapies and the importance of the idiographic approaches to SST.

The authors thank Kate Dazell, Jenna Jacob and Julian Edbrooke-Childs at Anna Freud CORC for their advice and recommendations on the Instrument. Dr Lynne Green and Dr Hannah Wilson from Kooth clinical team to safeguard the implementation of the project in the service. Tom Kayll to provide the data extraction and database from the service. Product team lead by Charlie Patterson to help the implementation of the instrument at Kooth. The team of practitioners at Kooth who review and safeguard every single free-text goal completed during the study. Lastly, thanks to Aaron Sefi and Kooth advisory board who enabled the funding for this study.

This project was received a favourable ethical review by the University of Manchester Research Ethics Committee – Ref: 2024-20027-34897.