The gap between research and practice

If all participants were aware of early intervention programs, most of them did not make use of predictive scales such as the CAARMS or the SIPS [19], nor did they refer their patients to early management centers. Actually, it would not make sense for all practitioners to use them, as these scales require extensive training. While ARMS assessment instruments do not predict psychosis in an individual, our participants think that they have the ambition to do so, and criticize it. Thus, they are reluctant to refer their patients to early management centers where ARMS might be assessed. However, either trained or not, all participants provided their young patients with multiple months follow up without use of an At Risk Mental State category (ARMS) [20,21].

Impossible to predict psychosis in daily work

Three sources of uncertainty

Psychiatrists’ impossibility to predict psychosis emerges from three sources of uncertainty.

The challenge of individual psychosis prediction

The first source of uncertainty mentioned by participants refers to medical judgment and is consistent with the sociological analysis of medicine. Fox argued that learning to cope with the uncertainty of medical practice is a professional norm that is assimilated early in the curriculum. She distinguished three sources of uncertainty: the limits of a physician’s knowledge, the limits of medical science, and the inability of the physician to distinguish between those two sources of uncertainty [25,26]. Freidson and Hunter argued that medicine is taught as a science of individuals. Each clinical case is unique and therefore uncertain [27,28]. According to Light, medical trainees learn to control uncertainty through specialization, involvement in schools of thought, clinical experience, and growing autonomy [29]. Thus, claiming to belong to a school of thought such as psychoanalysis or youth psychiatry could merely be way to cope with uncertainty. Dealing with serious diseases, prognosis uncertainty can be functional, that is used by doctors in order to avoid distress reactions from patients and their relatives [30] and avoid being held responsible for the patient’s outcome [8]. In that sense, a prognosis disclosure is a too risky commitment. This opinion is expected to persist over some period of time, to account for the psychiatrist’s “consistent behavior”. According to Becker, commitments come into being when a person, by making a side bet, links extraneous interests with a consistent line of activity [31]. Side bets are often a consequence of the person’s participation in social organizations. In daily work, a person sometimes finds that he has made side bets constraining his present activity because the existence of general cultural expectations provides penalties for those who violate them [31]. For instance, the cultural expectation about the behavior accounting for a practitioner’s trustworthiness is different from the expectation accounting for the trustworthiness of a researcher. People feel that a practitioner ought not to change his opinion too often and that one who does is inexperienced and untrustworthy [8]. On the contrary, a researcher’s shifting views on a subject might account for his commitment through his consistency in improving his study design. Thus, a practitioner will avoid prognosis disclosure, instead of sharing his worries about a psychosis onset to a young patient and then changing his opinion after a second appointment. Functional uncertainty is thus used by doctors in order to diminish their own level of commitment.

The second source of uncertainty mentioned by the participants to this study was the immediacy and unpredictability of teenagers’ symptoms. The adolescent period is seen as a social, hormonal, neurological, and psychological storm both by public opinion and by health professionals [32,33]. As early as in 1958, Anna Freud famously suggested that calm teenagers may hide distressful emotions: “being normal during the adolescence period is abnormal” [34]. The participants in our study argued that they were not always able to distinguish adolescent unease or rebellion from subdued psychotic symptoms. Interviewees’ vision of adolescence as an unpredictable process combining daily mood instability and social skills development is supported by neurosciences [35,36]. Teenagers new access to abstract thinking, if formerly stressed by child development psychology [37] is now supported by brain imagery studies revealing a neuronal network reconfiguration through synaptic trimming [38] that extends into the second and third decades of life [39]. Another difficulty is related to the fact that the incidence of positive psychotic experiences in the general population is 100 times greater than the incidence of psychotic disorders [40]. Moreover, in the adolescent and young adult population, psychotic-like experiences are strongly correlated with distress, depression and poor functioning [41].

Participants’ third source of uncertainty was related to psychosis diagnosis and course. First, we asked questions focusing on psychosis but participants’ answers focus on schizophrenia (in French “schizophrénie”) instead of psychosis (psychose). According to the Grounded Theory, this is neither a methodological problem, nor a translation mistake, but a meaningful result. The use of schizophrenia as a synonym for psychosis enlightens participants’ contemporary representation of psychosis as an equivalent of schizophrenia. The schizophrenia label, if still used as a reliable research category [42,43] is increasingly contested in order to reduce stigma [44] and account for a complex and poorly understood etiology [45,46]. The Ultra High Risk of psychosis (UHR) low prediction accuracy (15% to 25%) [47], high rates of false positive identifications (from 50% to 84%) [48,49] and difficulties to evaluate if early treatments improve long-term outcomes are stressed as current research challenges [50–54]. Research teams believe that the risk of psychosis onset during adolescence is far lower than 30% when SIPS or CARMS is applied in the general community [49]. Authors attribute this low predictive value to low prevalence of the psychosis risk syndrome in a general adolescent clinical population [55]. Thus, given the potential stigma associated to a Psychosis Risk syndrome [56] and its need to be further validated, it was not included in the DSM-V [57]. Second, participants expressed a contradiction between the practice of a probabilistic judgment in early diagnosis, due to the fact that psychosis diagnosis is based on a retrospective duration criteria [58]. Third, our results show that the vision of psychosis as a disastrous life-long trouble [59,60] refrains psychiatrists’ involvement in preventive approaches. Despite activism by former patients and patient advocates who have argued that psychosis did not prevent persons to live meaningful lives and to maintain a social role in their community, as well as scientific evidences that the course of schizophrenia is not always negative [9–12,61,62], schizophrenia’s major economic costs and duration shapes the idea that “the onset of psychotic illness represents potential personal disaster in the life-course of an individual” [63]. Thus, interviewed psychiatrists extended the deadline before disclosing such a dreadful perspective. However, if the primary outcome of interest in at-risk research was the development of psychotic disorder—whereby patients are categorized as either having ‘transitioned’ or ‘not transitioned’–recent research suggest that ‘non transitioned’ persons may have a poorer quality of life [64,65]. More than attenuated psychotic symptoms (APS), social and functioning difficulties as well as depressive symptoms can be the highest sources of distress leading UHR patients to seek help [66]. Thus, the term ‘outcome’ might be broadened to incorporate non-psychotic diagnoses, functioning and negative symptoms [67].

Health systems beyond the research/clinic dichotomy

Interactionism in sociology focuses on the way that people act and on the reasons they give to justify their action [68]. In this perspective, a meaning does not need to be “true” to be effective. The meaning that people give to a thing or an event determines its symbolic reality, which has the power to induce a social effect.

Which reasons can be used by our participants to support their views? Even if they know statistical data related to psychosis onset, in the end psychiatrists make decisions according to their clinical sense. This “in-between” strategy including both rational and non-rational strategies has been proven to be efficient in risk-management strategies [69]. Our results show that participants do not expect that early diagnosis tools improve the actual care provided. They legitimize their reluctance toward ARMS assessment by the clinic/research dichotomy. For instance, two reasons are used by our participants to avoid the referral of their young patients to assessment centers: first, it is impossible to predict an individual outcome; second, it is impossible to know if psychosis can be prevented in an individual. However, researchers of the ARMS field already agree with those critics [49]: first, ARMS assessment instruments aim at determining an ARMS, not predicting psychosis in an individual. Second, early interventions for ARMS individuals aim at improving functional outcome at group level. Nevertheless, “valid” or not in a scientific meaning, these reasons induce a real social effect: our participants’ reluctance to refer their young patients for ARMS standardized assessment. According to the Grounded Theory, this confusion does not reveal a methodological mistake or participants’ lack of perspective, but a meaningful result. Moreover, we will argue in the following section that participants’ inconspicuous risk management might as well be understood considering the characteristics of the French health system.

How do our participants actually act? When they suspect a psychosis onset, interviewees start an inconspicuous risk management: multiple months follow up of this fuzzy clinical state. We suggest that this strategy may be influenced by the organization of youth mental health services in France. In this country 77% of mental health care is provided by the public system with a universal costs coverage [70]. A diagnosis of risk is not necessary in order to obtain reimbursement of mental health costs. Thus, psychiatrists use their subjective judgment to start a discreet follow-up. As early as in 1941, Rümke described the “praecox feeling” as a feeling of unease emanating in the interviewer that reflects the detachment of the patient and the failure of an ‘affective exchange’ [71]. Praecox feeling reliability has mainly been evaluated among persons with an ongoing psychosis. Compared with the standardized diagnostic classification the precision of the intuitive reasoning appears remarkably high [72–74]. On the contrary, in a U.H.R. population, Nelson and Yung showed that the clinical impression of an incipient psychosis—that is, a clinical judgment that a particular patient may soon progress to a full-threshold disorder—is not sufficient for predicting a psychosis outcome and may lead physicians to improper antipsychotic medication [75]. This result stresses the need to design more adequate follow ups for young people. Youth psychiatry (child and adolescent) services usually stop following patients when they reach 18 years of age and are then referred to adult services. The communication between those services remains limited both for clinical and for research purposes. Today, despite psychosis peak onset in late adolescence—early adulthood [76], only few youth psychiatry departments are designed for following up young patients until 25 to 30 years. The participants to our study usually did not receive news from their patients after they had been referred. The lack of outcome studies following young patients from child to adult departments and of services specially designed for persons aged 15–30, increases the youth psychiatrists’ challenge to imagine young patients’ evolution after their 18s.

Study limitations and implications: Rationale for a comparison of local mental health system changes

If qualitative methods’ validity is increasingly acknowledged in medicine, the Grounded Theory (GT) remains rarely known in the medical field. If Interpretative Phenomenological Analysis (IPA) is the dominant method increasingly used in psychological studies—and previously employed by our team [77]–its’ main difficulty is to develop the analysis to a sufficient interpretative level [78]. IPA aims to understand the experiences (phenomenology) and meaning-making activities (interpretation) of the research participant [18]. The analyst attempts to make sense of the participants’ interpretation of their own experience, thus creating a double hermeneutic [79]. Nevertheless, improving the interpretation level can be facilitated by borrowing techniques from the GT [80]. Beyond the individual psychological experience described by IPA, the GT emphasizes its particular societal, cultural and political context. That is, participants’ views convey some knowledge about beliefs and practices shaped by professional norms and institutions’ organization. IPA is better known than GT in the medical field due to the historical fact that most mental health qualitative research has been designed by psychologists in order to grasp individual experiences [81]. In the present context of positive mental health, Grounded Theory provides a particularly relevant approach for qualitative researches assessing the local impact of contemporary health system changes.

The sampling strategy was to encounter diverse range of youth psychiatrists working in different settings and being either actively involved in research or not [17]. However, if psychiatrists had distinctive clinical practices and attitudes toward research, they were unanimous about their impossibility to commit in prognostications in their daily work. Thus, saturation was rapidly reached. A validity criterion of qualitative studies is to expose results of the analysis to participants. Findings were directly communicated to four participants and five oral communications were addressed to more than 70 youth psychiatrists’ who agreed with our results (oral communications in the department meetings of three hospitals Cochin, Pitié-Salpêtrière, and Avicenne (Paris Region), the annual conference of de French Society of Child and Adolescent Psychiatry (SFPEADA) in Lyon, May 2016, the annual conference of French Society for Psychiatric Information (SIP) in Bruxelles, September 2016). While it is not possible to generalize from such a small scale study, the opinion reported here may reflect the perspectives of a significant segment of French youth psychiatrists. This study is first step to provide an account of early interventions implementation. The W.H.O. argues that given their historical and conceptual diversity, the efficiency of the mental health policies and practices in different countries is difficult to compare in an evidence based perspective [82]. Using the GT in different countries would offer a qualitative comparison of young people access to mental health care and a particularly relevant insight of psychosis early management implementation in psychiatric services.