Developing patient profiles

The patient profiles were developed at Lovisenberg Diaconal Hospital’s SUD outpatient clinic, whose catchment area is the eastern central districts of Oslo. The catchment area is characterised by a concentration of young adults and a large variation in social functioning. All patients were eligible for treatment in the SUD outpatient clinic, typically indicating moderate to severe SUD and comorbid mental disorders excluding severe mental illness.

The approach we used in developing patient profiles shares some similarities with grounded theory (GT) (Glaser & Strauss, 1967), which is a qualitative method that describes how novel theoretical frameworks are developed through systematic comparison of different forms of empirical data. We developed the patient profiles by comparing patients we perceived as similar. GT provides useful methodological insights into such work, especially in relation to the consistent extraction of profile content from comparisons of specific patients. However, the purpose of our study was to gather specific characteristics into a rich, experience-near description of individual patients with a view to facilitating the sharing of experiences, whereas in a standard GT study, the aim would be to develop precise and general concepts to explain empirical data as part of a theory.

The patient profiles were developed in a four-step process that took place over a five-week period.

In step one, Sigurd Syrdal Aanderaa, Linn Bjerknes and Mette Nordbrønd Mikkelsen started by individually responding to an open instruction: ‘Divide your patient portfolio into five groups based on which patients you perceive to be similar.’ To obtain a tangible representation of the patients, we all wrote the initials of the individual patients on Post-it notes. We experimented by placing the notes in different groupings. During this process, we identified certain aspects that could lead to the perception of similarity. One of us noticed, for example, that all the patients placed at one side had no treatment plan.

In the next step, we presented the groups to each other. The preliminary groups in this phase were described using keywords characterised by obvious characteristics. For example, a category could be described in the following terms: early extensive substance use, been involved with the child welfare authorities, home for adolescents, personality issues, has had or currently needs inpatient treatment, treatment fatigue, problems with relating to the therapist. We identified a number of implicit dimensions that guided the grouping process, including the perception of: difficult to work with, relational stability, level of functioning, degree of insight, common treatment project. We noticed that some of the groups were similar. The groups with similarities served as inspiration for the development of patient profiles.

In the third step, we elaborated on the categories by comparing the descriptions of each patient with similarities. We then added specific patient experiences that we felt resonated with the description. In shaping the profiles, we jointly reviewed the proposed categories in several rounds in order to anchor them in patient experiences. Our discussions led to three profile outlines that we wanted to work on further. The three-profile limit was related to the timeframe of the project. In the process of identifying and shaping the profiles, we wanted to alternate between individual reflection and group work in order to access our own intuitive understanding without interference. 

In the fourth step, we chose to organise the patient profiles in a prototype format (Westen, 2012), where each of us drafted a coherent description of the three profiles. We wanted to include a wide range of clinically significant factors and opted to base them on information typically gathered in an initial session: about substance use and history, psychological symptoms and mental disorders, level of functioning, childhood, adolescence and education, social networks and labour market attachment. We subsequently synthesised the three drafts of each of the profiles into a unified description. 

We then tested the extent to which we felt the patients in our own patient portfolio matched the three patient profiles. Using a scale developed by Westen (2012) for prototype-based classification, we scored each patient based on how well they matched each of the three profiles according to the therapist’s perception of similarity between the aprofile and the overall impression of each individual patient. The scale ranges from 1 to 5, where 5 indicates that the profile describes the relevant patient to a very large extent, 3 to a significant extent and 1 not at all. Then, as a group, we made a more in-depth comparison, where we explored the rationale behind the scores each of us had given.

We also tested out the use of patient profiles for sharing experiences with our group of colleagues. Fourteen therapists at the SUD outpatient clinic scored their own patients in relation to the patient profiles. For each profile, we shared experiences regarding typical challenges and useful strategies in our clinical practice. 

Stereotypical C profile

The first, and most comprehensive finding concerned patient profile C. We all had a perception of a group of patients who were similar to each other in that they smoked cannabis. These were often young men who had little insight into how their cannabis use was preventing them from progressing in life. It transpired that several patients with cannabis as their primary substance had received high C scores, despite the overall clinical picture for some being more similar to profile B, with good social functioning and greater insight into their own substance problem. It also emerged that patients whose overall pattern matched profile C, with social and psychological functioning characterised by withdrawal and passivity, but did not have cannabis as their primary substance, were not identified as a C profile. The review highlighted how patient profile C was influenced by our preconceptions of a ‘typical cannabis patient’, and that characteristics of the profile were taken directly from these preconceptions. These preconceptions, where we linked a pattern in psychological and social functioning to the use of a specific substance, i.e. cannabis, influenced the development of patient profile C. By comparing profile C with the experience of specific patients, it became apparent to us that our perception of the typical cannabis patient was a stereotype. We found patients with cannabis as their primary substance who were not characterised by lack of insight and passivity, as well as patients with alcohol as their primary substance but who otherwise closely matched the profile. 

Negative weighting in the A profile

The second main finding related to us identifying fewer of our own patients as a match for the A profile than the patients of the other two clinicians. In other words, when discussing each other’s patients, we were more likely to consider the other clinicians’ patients to be a better match with the description than our own. We became aware that the profile had an unintended negative weighting. We all found the profile description to be more unequivocally negative than our perception of our own patients. The profile lacks more unequivocally positive qualities, such as the ability to show vulnerability and be direct in relationships. We tend to like and sympathise with our own A-profile patients. In retrospect, we have considered that this represents a ‘split’ in the perception of patients. When therapists encounter patients who fit the description of profile A, they often have difficulty reconciling the different experiences they have with those patients into a cohesive understanding. 

Patient profiles versus case discussions

Discussing cases is invaluable both for providing input and for collective learning within a collegial community. However, the discussions can be quite time-consuming, which means that priority is given to patient cases that present particular challenges. This leads to a bias in relation to which patient cases are discussed, consequently reducing the opportunities to share our experiences with effective treatment in typical, less dramatic patient cases. The patient profiles were generated based on a comparison of all the patients on the lists of the three clinicians, and the resulting descriptions are representative of different segments of the patient population. We consider the three profiles we developed (A, B and C) to be very distinct from each other, thus enabling us to discuss a broader range of patient cases.

Presenting challenging patient cases for discussion can be emotionally taxing, and it is important for colleagues to adapt their input according to the capacity of the clinician who ‘owns’ the case. This sensitivity can limit the sharing of experiences by colleagues, for example success stories with similar cases to the one being discussed. We found that sharing experiences in the collegial community based on the patient profiles led to a more open and explicit comparison of experiences, and we found that we all had different ideas about what types of patients were most difficult to work with. During the discussion process, it became evident which strategies the clinicians perceived as useful for profiles A, B and C, and there was some disagreement about therapeutic strategies that we were not aware of. In contrast to case analysis, where the focus is to help provide a solution to a specific clinical problem, the profiles provide an opportunity for a more overarching systematic exchange of experiences.

Patient profiles versus diagnoses

A common feature of diagnoses and profiles is that they both represent categorisations that facilitate the development of explicit knowledge. Diagnoses can sometimes help the clinician to identify useful therapeutic approaches, not least because treatment research often uses diagnostic categories. For example, both a withdrawn middle-aged man receiving disability benefits and a young student who has recently lost control of their substance use can be diagnosed with alcohol dependence. In such cases, comparing treatment experiences based on diagnosis makes little sense. As part of the work on modern diagnostic systems, diagnoses have been developed by collecting data on intercorrelated symptoms (Blashfield et al., 2014). The aim was to arrive at diagnoses that corresponded to underlying syndromes, but this has not proven to be successful (Hyman, 2010; Kupfer et al., 2002). In contrast to diagnoses, the patient profiles presented here are based on the therapist’s intuitive and local knowledge of which patients appear to have similarities in terms of potential treatment. Through articulation and systematic comparison of intuitive knowledge, our aim was to group together patients who could benefit from similar approaches. The collegial community found that each profile had its own typical challenges, and that a consensus could be reached on a common set of useful strategies for each profile. 

We cannot automatically assume that the apparent similarities of patients represent valid intuitive knowledge. Klein (1998) understands experts’ spontaneous and efficient perceptions as an expression of intuitive knowledge. Experts are quick to identify relevant goals and measures, as well as which aspects of the situation are important to monitor, without being able to explain how they arrived at their understanding of the situation. However, valid intuitive expertise in a field requires sufficient experience with regularities over time (Kahneman & Klein, 2009). Whether clinical psychologists experience enough observable regularities to develop valid intuitive knowledge has been the subject of debate (Dawes, 1994; Kahneman, 2011). Kahneman (2011) argues that psychotherapists may possess valid intuitive knowledge about, for example, regulating anger or building trust, as they have ample opportunity to observe immediate reactions to regulating interventions. We assume that therapists can have valid intuitive knowledge about what types of goals can be achieved in a conversation with a patient and what can contribute to this, such as the patient’s ability to understand the correlations in their own life or the patient’s need for emotional validation. When experienced clinicians group patients according to similarities in terms of potential treatment, this will activate intuitive knowledge about how different patients need to be approached. Conversely, diagnoses used in actuarial assessments (Dawes, 1994) will be better suited for prognostic considerations over a longer period of time. 

Generalisability of patient profiles

Patient profiles are explicitly a local categorisation, in contrast to diagnoses, which are part of a universal classification system. One of the characteristics of intuitive knowledge is its context specificity; it is valid within the specific context in which it is developed (1998). Differences in catchment areas and admission criteria make for distinct patient populations, and the objectives and frameworks that play a role in determining treatment pathways will vary across the different types of units. The profiles of patients in a detoxification unit will also look quite different, even if some of these patients are the same. The patient profiles we develop at an SUD outpatient clinic in Oslo’s Inner East will not necessarily be representative of patients at such a clinic elsewhere in Norway. The method of developing typical patient profiles can be useful for a wide range of clinical units; however, the strength of scientifically applicable diagnostic categories is that they can be used as a basis for research into treatment methods for different groups of patients in a variety of settings. Patient profiles serve as a supplement to diagnostic categories, and they can help clinicians integrate evidence-based knowledge and enable them to share experiences in tailoring evidence-based interventions to different types of patients.

Patient profiles as a process tool

A key point in the method for developing typical patient profiles is that it can be carried out in a normal clinical context. We have used our own recollections and understandings of our patients as a form of ‘empirical data’ in the comparison. Glaser and Strauss (1967) discuss the use of personal recollections and also acknowledge this as a possible sub-strategy in GT. Time-consuming methods, such as semi-structured interviews and their subsequent transcription, are unrealistic if the patient profiles are to be developed by therapists in normal clinical settings. However, our use of ‘empirical data’ limits the opportunities for others to check or evaluate the comparisons on which the profiles are based. Nevertheless, the aim of the project is not to create universally valid categories but to describe steps in a method that can be used in different local contexts. The method has enabled us to correct our own stereotypes and misconceptions and to share experiences with a view to increasing the body of clinical knowledge in our clinic. 

We observed with both profile C and profile A that the methodical development of patient profiles was useful for alerting us to and challenging our own perceptions of patient groups. In the project, we understood the importance of identifying the characteristics of the profiles through systematic comparison of specific patients, jointly reviewing the proposed categories in several rounds, and checking how well the patient profiles matched specific patient experiences. If the profiles are not sufficiently anchored in specific patient experiences, there is a risk that they will be based on stereotypes. 

The method of grouping patients enables the articulation of tacit knowledge, leading to an awareness of implicit categories. One benefit of articulating intuitive groupings or categorisations is that it facilitates nuanced reflection on stereotypes, prejudices and biases that are inherent in any treatment culture. After developing and documenting our local patient profiles, we noticed that the patient descriptions lacked information about somatic diseases and conditions. Those of us working on the project are psychologists, and our SUD outpatient clinic has been dominated by psychologists and social work therapists, with a historically low proportion of doctors and nurses. The process of developing the patient profiles revealed the underrepresentation of the medical perspective, thus facilitating the identification of tacit assumptions in the treatment culture. This imbalance in representation also highlighted the importance of obtaining perspectives from the wide range of practitioners in shaping the profiles. 

Further development of the method

In the project, we devised a method for developing a typology that can facilitate the systematic sharing of experiences among clinicians in a collegial community, thereby fostering the development of local knowledge about effective treatment for different types of patients. We have not systematically investigated whether the patient profiles facilitate the sharing of experiences, but we have used the profiles in discussions between ourselves. Our observation to date indicates that, compared to case discussions, the profiles stimulate more systematic treatment discussions. Going forward, the categorisation method and patient profiles should be further developed and evaluated. We envisage more direct investigations into whether patient profiles facilitate the comparison of treatment experiences. It will also be important to develop patient profiles in other clinical units in order to gain experience with the method and the range of different local patient profiles in practice.