Examining service‐user perspectives for the development of a good outcome checklist for individuals at clinical high risk for psychosis

Around 15% of patients at clinical high risk for psychosis (CHR‐P) experience symptomatic remission and functional recovery at follow‐up, yet the definition of a good outcome (GO) in this population requires further development. Outcomes are typically designed and rated by clinicians rather than patients, to measure adverse as opposed to GOs. Here we investigate how CHR‐P subjects define a GO, with the aim of developing a checklist that could be used to measure GO in this clinical group.


| INTRODUCTION
Traditionally, clinical outcomes in individuals at clinical high risk for psychosis (CHR-P) (Fusar-Poli, 2017) have been defined with respect to transition to psychosis, however, recently, outcomes have also been described in terms of symptom remission and functional recovery (Yung & Nelson, 2013). The onset of psychosis is a clinically meaningful outcome, as it usually necessitates a change in clinical management and treatment. Similarly, if presenting symptoms improve and an individual no longer meets criteria for the CHR-P state, they may require less clinical support. Assessing functional outcome is also useful as some CHR-P subjects who do not develop psychosis continue to function poorly in comparison to the general population (Addington et al., 2011). Conversely, some CHR-P patients who become psychotic retain relatively good functioning. Therefore, level of functioning may provide a measure of disability independent of symptom-severity or transition status (Carrion et al., 2013).
Around 15% of the CHR-P population exhibit symptomatic remission at follow-up, they are not psychotic, and are functioning well . The CHR-P group exhibit several longitudinal clinical and functional trajectories (Allswede et al., 2020;Polari et al., 2018), including outcomes such as remission and recovery.
However, a standardized and well-established definition of a favourable outcome in this cohort is still in development, particularly one that incorporates multiple domains (eg, symptoms, functioning, comorbidities and personal wellbeing). We conducted a Delphi study with a panel of CHR-P expert researchers and clinicians (n = 46) to establish consensus on factors that could be used to define a 'good outcome' (GO), and to develop a framework for assessing GO in CHR-P patients (Petros et al., 2019). The experts defined GO in terms of a combination of clinical, functional and wellbeing metrics, consistent with the multi-faceted nature of the CHR-P state . Ninety-eight items were endorsed by ≥80% of the sample, and 30 were considered significant for defining short-and long-term GO. These fell into four major domains: Functioning; Symptoms; Distress and Suicidality; and Subjective Wellbeing. A good level of daily functioning was regarded as fundamental for indicating GO, and was rated above role and social functioning, and symptom amelioration. The largest domain, containing the greatest number of GO indicators, was Subjective Wellbeing, representing personally meaningful, service-user-defined outcomes (eg, personal goalachievement), typically regarded as features of 'recovery' (Jacob, 2015).
This suggests that, ideally, a combination of patient and clinician-rated items should be used to evaluate GO in CHR-P patients.
Whilst conventional measures of outcome in the CHR-P population are useful, they were originally developed to determine adverse outcomes, rated by clinicians not patients. Identifying the views of patients with lived experience in outcome-focused research supports the ongoing movement towards mental health care centred on client values and preferences (Byrne, Davies, & Morrison, 2010). To our knowledge, this is the first study to explore which GOs are meaningful to the CHR-P population. The results complement data from our previous study (Petros et al., 2019) and will inform the development of a CHR-P-specific GO checklist.

| Data analysis
Data were analysed using SPSS (Version 25 IBM Corporation 2017).
Demographic characteristics were described using frequencies, percentages, means and SDs. Responses for each GO indicator were analysed using the relative importance index [RII (Holt, 2014)] method, with the formula: RII = ΣW AÃN , where W was the number of participants selecting each response on the Likert scale multiplied by the response point integer (ranging from 1 to 5), A was the highest weight (ie, 5) and N was the total number of participants. The GO indicators were ranked for importance using RII, the higher the RII value, the more important the indicator was for GO. The ranked data were then examined using percentiles. To complement the RII analysis, the cumulative percentage of endorsement for the GO indicators was calculated, a method comparable to the analysis used in Petros et al. (2019).
Items were considered important indicators of GO if ≥80% of the sample 'agreed' or 'strongly agreed' with the GO statement.
The GO items were separated into domains (Table 1) that overlapped with those that emerged from our previous work (Petros et al., 2019). Domain categories were determined following independent rating (by NP; AEC and SV) with disagreements resolved by discussion. This process led to the inclusion of three additional domains (5, 6, and 7) resulting in seven GO domains. Mean scores were calculated for each GO domain (sum of Likert responses/number of items).
Domain scores and total psychometric scores were assessed for normality using histograms. Spearman's correlation analyses were performed to examine associations between the GO and protective factor domains and between the psychometric and domain scores.
Due to the exploratory nature of these analyses, statistical threshold of P ≤ .01 was employed.
The majority (75%) of patients reported an improvement in their mental health and wellbeing subsequent to presentation, and 83% stated that they felt better at the time of assessment. However, over half of the sample exhibited mild-to-moderate depressive symptoms, anxiety and moderate-to-severe impairment in role and social functioning, and half reported moderate-to-severe psychological distress (Table 2). The sample demonstrated low wellbeing and recovery scores, with a mean recovery score comparable to that seen in patients with chronic psychosis (Slade et al., 2015). The mean number of different psychotic-like experiences (PLEs) reported was 6.46 (±3.96); 93.8% of patients reported at least one PLE, with 10% reporting more than 12 in the 3 months preceding assessment.
The GO indicators and protective factors were grouped into domains. Spearman's correlations revealed inconsistencies in the relationship between the GO domains (Supporting Information Material 2); this was less true for the protective factor domains, suggesting minimal floor or ceiling effects.  Table 1).

| Relative importance index
Patients indicated that items within the Subjective Wellbeing domain were most protective of their mental health; they also considered support from mental health services as having a vital influence on their improvement (Table 4)   Some patients met criteria for more than one CHR-P subgroup, multiple diagnoses and were prescribed more than one medication. b Subtherapeutic dose for psychosis.

T A B L E 3
Top 25% of items ranked as important for short-term and long-term good outcome from the perspective of service-users at clinical high risk of psychosis by experts (Petros et al., 2019), were not ranked highly by patients.
Improvement of non-specific and negative symptoms were deemed important. We have used data from our two investigations to develop a CHR-P-specific GO checklist.
Both CHR-P patients and professionals (Petros et al., 2019) regarded factors relating to Subjective Wellbeing as important for indicating GO. Such items have been linked to recovery (Jacob, 2015) in patients with established psychosis (Slade & Hayward, 2007). In the present study, patients indicated that learning from the experience of mental health problems was crucial to indicating short-term GO. This notion of growth post-adversity is an inherent feature of resilience (Fletcher & Sarkar, 2013), a construct rated by our sample as being critical for long-term GO. Resilience can be seen as a process of adaptation in the face of adversity (Bonanno, 2004), and likely plays a role in enabling people to maintain wellbeing despite susceptibility to mental illness. There are few CHR-P studies on resilience, however, findings suggest that low baseline resilience is associated with poor psychosocial functioning and transition to psychosis (Kim et al., 2013).
To enhance mental wellbeing and functioning in the CHR-P population, psychological interventions could be developed to promote resilience, in addition to reducing symptoms, our patients supported this notion by rating protective factors highly if they centred on selfreliance and personal strength (eg, self-efficacy/agency). A good working alliance with psychiatric services was also considered important for fostering GO. This finding supports research indicating a link between established therapeutic relationships and service engagement in early-psychosis patients (Browne, Nagendra, Kurtz, Berry, & Penn, 2019;Lecomte et al., 2008), which may set the foundation for successful clinical intervention (Svensson & Hansson, 1999). Surprisingly, support from friends/family was not considered essential for GO by the sample, despite being a known protective factor for common mental disorders (Smyth, Siriwardhana, Hotopf, & Hatch, 2015), and being associated with better functioning in psychosis patients (Pruessner, Iyer, Faridi, Joober, & Malla, 2011).
A reduction in distress and fear associated with symptoms was considered important for indicating short-term GO by the sample and was rated higher than symptom reduction and remission, consistent with evidence that distress, not attenuated psychotic symptoms (APS), motivates CHR-P patients to seek professional help (Falkenberg et al., 2015). Improvement of 'non-specific' presenting symptoms was rated highly for short-term GO; however, responses were dependent on what patients believed their presenting symptoms were (ie, psychotic or features of comorbid axis I disorders).
Comorbidities are common in CHR-P patients (Addington et al., 2017), therefore, we asked about APS explicitly. We could not exclude the possibility that patients did not believe their symptoms were psychotic, thus, the item on presenting symptoms was included to capture opinion on what prompted their help-seeking, without being too prescriptive. These 'non-specific' symptoms could have been viewed as psychotic by some patients, whilst for others they may have been viewed as mood-or anxiety-related.
Negative symptom amelioration was deemed important for long-term GO by the sample, as we previously found in professionals (Petros et al., 2019). This suggests that negative symptoms should be given greater priority in the assessment and monitoring of CHR-P patients, particularly as they are associated with psychosocial dysfunction (Kim et al., 2013) and poor outcome (Salokangas et al., 2014). Furthermore, negative symptoms are associated with poor resilience in CHR-P patients (Marulanda & Addington, 2016), and could impact the development of secure interpersonal relationships, a key ingredient for building and maintaining resilience (Bretherton, 1996).
The sample did not consider good functioning (daily, social or role) as being highly important for indicating GO, unlike the professionals (Petros et al., 2019), moreover, good functioning was considered less T A B L E 4 Top 25% of protect factor items ranked as important for yielding good outcome from the perspective of service-users at clinical high risk of psychosis important in the more symptomatic patients (eg, more depressed). Our findings are inconsistent with reports that functioning may be a better indicator of CHR-P outcome than symptomatic status (Brandizzi et al., 2015;Carrion et al., 2013). It is important to note that the overall result could be a reflection of symptom-severity within the group.
Depressed mood in CHR-P patients has been linked to motivational deficits, particularly in relation to anticipatory pleasure (Schlosser et al., 2014), which could lead to social isolation and avoidance of vocational goals. Therefore, the low ratings for 'good functioning' as a GO indicator could signify difficulties in symptom-related motivation in our sample, which also exhibited poor functioning.

| CHR-P GO checklist: Design, rating and utilization
At present the method for assessing GO in CHR-P patients requires further development; we extended our initial framework ( The GO-12 incorporates items identified in our previous (Petros et al., 2019) and current work as being important to CHR-P professionals and patients. We have included items related to functioning based upon the high ratings given by CHR-P-experts and in light of evidence suggesting the importance of functioning as an outcome indicator (Brandizzi et al., 2015;Carrion et al., 2013). Some symptomrelated items were split between short-term (<1 year) and long-term (>1 year) GO, negative and positive symptoms were included, with focus on amelioration in the short-term and remission in the longterm. Improvement of presenting symptoms would be difficult to measure reliably, as these symptoms are 'non-specific' and different for each patient, therefore could be assessed qualitatively. A reduction in distress associated with symptoms was important to both groups and a reduction in fear of symptoms was deemed essential to patients. Resilience was one of the most important indicators of GO to patients and is linked to tolerance to stress; an item rated highly by professionals (Petros et al., 2019).
The scoring system of the GO-12 checklist was inspired by the Historical-Clinical-Risk Management-20, Version 3 (HCR-20V3) (Douglas, 2014), a comprehensive risk assessment used in adult forensic settings. The aim is for each GO-12 item to be scored based upon presence, with a higher total score indicating better outcome. The checklist employs a simple, yet comprehensive design, allowing for integration into routine assessment using data from clinician-rated tools and patient-reported outcome measures. For the checklist to be universal and accessible, data collection methods are not restricted to specific instruments or parameters. Several widely used instruments that have demonstrated validity and reliability in the CHR-P F I G U R E 1 The good outcome 12-item (GO-12) checklist for patients at clinical high risk for psychosis population could be used to determine GO for each item. For example, the Comprehensive Assessment of At-Risk Mental State (Yung, Phillips, Yuen, & McGorry, 2006); Social and Occupational Functioning Assessment Scale (Goldman, Skodol, & Lave, 1992) and Resilience Scale for Adults (Friborg, Hjemdal, Rosenvinge, & Martinussen, 2003) could be used to assess symptoms, functioning and resilience, respectively. The next phase of our work will involve assessing the validity and reliability of the GO-12 using pre-existing datasets and assessing the feasibility of scoring the GO-12 using data collected by different instruments.
A major limitation of this study was the modest sample size; however, the number of patients involved matched the number of professionals that participated in our previous study. Participants were self-selected, this was a potential source of bias, however, those who participated were demographically similar to those who did not.
A further bias relates to treatment experience, most of the participants had received psychological intervention from their care team (Table 2), therefore, it is possible that specific items, for example, 'access to psychotherapy' may not have ranked as highly in a medication-only or no-treatment sample. The cross-sectional design of the study precluded a longitudinal assessment of the stability of the patients' views, which may have differed pre-and post-clinical intervention.
This is the first study to examine GO in the CHR-P group from the perspective of the patients and to develop a checklist that can be used to assess GO. Our findings indicated that CHR-P patients have differing views on what constitutes a GO compared to professionals, suggesting that a collaborative effort is key to the development of measures that assess outcomes in this population.