Making a diagnosis of a particular mental disorder in children and adolescents can be a challenging or overwhelming task for clinicians or researchers who are less experienced or have to complete their assessment within a limited time frame. Such an evaluation can seem impossible, especially if a clinician intends to make a well-supported DSM-5 [1] diagnosis. Each step of a DSM-5 based diagnostic procedure with a child and his or her caregivers (i.e., examining if a child’s clinical presentation fulfills the specific symptom-based diagnostic criteria, judging if the symptoms are not caused by other diagnoses or stressors and evaluating the impairment caused by these symptoms on a child’s functioning) necessitates a practical use of time to gather appropriate information [2].

The categorical model of mental illness preserved by DSM-5, in which a person does or does not have a mental illness, is widely recognized for its diagnostic reliability (i.e., the ability of different clinicians to agree on the diagnosis of a particular mental disorder for a particular person) [3]. However, diagnostic validity of the categorical model (i.e. the ability of clinicians to make an accurate diagnosis) is considered limited [3, 4]. The dimensional model aims to improve the variability and validity of psychiatric diagnoses through measures that quantify and personalize the diagnostic criteria for a particular person according to his or her own subjective experience of core symptoms and with reference to normative data of clinical or community samples. These measures include short symptom statements that a child or his/her caregiver assesses, often with a five-point scale of symptom levels. Beyond identification of presenting problems, these measures have the potential benefit of assessing treatment response and progress toward recovery through measurable outcomes. However, these measures are not designed to indicate the presence or absence of a specific diagnosis since they assess dimensions of symptomatology rather than fulfillment of diagnostic criteria [5, 6].

As far as child and adolescent mental health assessment is concerned, an additional debate is whether children provide reliable or valid clinical information during diagnostic interviews, even though researchers and clinicians generally agree that it is beneficial to hear a child's account of his or her presenting problems. In fact, latest research has documented that children can contribute relevant, clinically useful, valid information during clinical psychological assessments [7].

In the context of these challenges and limitations, we developed a fully structured interview in the Greek language, the Computer-Assisted Structured Diagnostic Interview (CASDI), which encompasses categorical and dimensional approaches and automatically navigates the interviewer, through computerized decision trees (algorithms), in a specified probing system including: (a) screening questions regarding severity of most common symptoms during the last 12 months, (b) follow-up questions measuring impairment due to reported common symptoms of at least mild severity and, when appropriate, (c) symptom questions regarding severity and duration in the diagnostic criteria of child and adolescent mental health disorders. A self-response interview (CASDI 8-18) for children and adolescents aged 8-18 years and a proxy interview (CASDI 6-18) for caregivers of youths aged 6-18 years are available. The duration of the interview with each informant (child or caregiver) can vary from 30 minutes (subject with no psychiatric diagnosis) to 180 minutes (subject with three comorbid diagnoses).

After an initial five-minute phase of introducing interviewer to the informant and listening to the informant’s uninterrupted speech about reason of referral, the interview proceeds to 15 open-ended questions about the history of the current illness, with a focus on the “why now?” of the presentation and precipitating events, psychiatric history of the child, past medical history, family history, developmental and social history. Beyond valuable qualitative information that is yielded in the above phases of the interview, this time builds the alliance and allows the interviewer to understand much of the mental status of the child. Also, 12 questions regarding child’s functional impairment elicit a response from the informant in a five-point rating scale of impairment level (None-Slight-Mild-Moderate-Severe). Additionally, in the proxy interview, a total of 21 questions assessing symptoms of specific learning disorders is also included. This section is expected to be completed in 10-15 minutes.

A total of 343 screening questions, follow-up impairment questions and severity/duration symptom questions for each of the DSM-5 categories of mental disorders-experienced by children and adolescents and reported by themselves or their caregivers-follow next. This set of questions constitutes the core diagnostic section of the CASDI. All questions intend to elicit a response from the informant in a five-point rating scale of symptom severity or impairment levels (None-Slight-Mild-Moderate-Severe), with the exception of responses to questions measuring symptom duration that are coded categorically. Information about diagnostic criteria that are observed rather than elicited, as in the case of disorganized speech, is categorically coded based on interviewer’s responses (Yes-No). Additionally, the interviewer reports (Yes-No-Unsure) if any symptoms are attributable to another medical condition or to the physiological effects of a substance/medication. Diagnostic categories and specific diagnoses reviewed through the CASDI questions are presented in Table 1 with reference to DSM-5 and respective ICD-10 codes. In case that criteria for the abovementioned diagnoses are not fully met, the CASDI prompts the interviewer to consider alternative diagnoses, mainly either unspecified/provisional disorders that need to be excluded or reassessed in a later time or diagnoses that require more systematic assessment, such as individualized, standardized intelligence testing or laboratory tests. These alternative diagnoses are presented in Table 2.

The interview with the child or the caregiver ended with a two-minute set of follow-up open-ended questions asking whether the CASDI questions addressed informant’s major concerns, if there is anything important the CASDI missed or anything that the interviewer should know about to better understand what the informant is going through.

The last section of the CASDI includes a set of questions for interviewer-rating only to be made after the completion of interviewing the child or the caregiver. Findings from the mental status examination regarding child’s appearance, behavior, speech, emotion, thought process, thought content, cognition and intellectual resources, and judgement and insight are categorically (Yes-No) coded here. Last, the interviewer is asked to mark, if appropriate, the presence of any conditions and problems from a selected list provided by the DSM-5 [1]. These conditions include mainly psychosocial and environmental problems that may be a focus of clinical attention or that may otherwise affect the diagnosis, course, prognosis, or treatment of a child’s mental disorder. Coding these conditions and problems aims to draw attention to additional issues that are encountered in clinical practice and it is useful to be documented by clinicians.

Separate scores and diagnoses by child and parental reports are generated. The CASDI automatically informs interviewer during online completion about the mental disorders (coded according to ICD-10) for which the DSM-5 diagnostic criteria are fulfilled, based on children's and parents' reports. If a diagnosis is reported as being present by either the child or the parent, then it is counted as being present for the purpose of making a joint diagnosis.

The present study aimed to assess concurrent validity of the CASDI, using as a gold standard the clinical diagnoses formulated using the DSM-5 diagnostic criteria and based on the DSM-5 Pediatric Diagnostic Interview [8].

A total of 258 (i.e. 90.53% recruitment rate) children and adolescents (46.51% boys) aged 8-18 years (Mean±SD: 12.66±2.73) and their parents completed the CASDI. Close to half the sample had an intermediate level of family affluence (45.35%), while 36.43% had low level and 18.22% had high level of family affluence.

The CASDI generated in average 1.5 diagnoses per participating child, while the member of the team (an experienced child and adolescent psychiatrist) who conducted DSM-5 Pediatric Diagnostic Interview formulated 1.2 diagnoses per subject.

The validity indicators of the CASDI in diagnosing specific 12-month DSM-5 diagnostic categories are shown in Table 3. The CASDI had sensitivity above 70% for Neurodevelopmental Disorders, Schizophrenia Spectrum and Other Psychotic Disorders, Depressive Disorders, Anxiety Disorders, Obsessive-Compulsive and Related Disorders, Trauma- and Stressor-Related Disorders, Somatic Symptom and Related Disorders, Feeding and Eating Disorders, Elimination Disorders, Sleep-Wake Disorders, Gender Dysphoria, Disruptive, Impulse-Control, and Conduct Disorders, Substance-Related and Addictive Disorders and below 50% for Bipolar and Related Disorders and Dissociative Disorders.

The CASDI was well accepted by both children and parents and there were no refusals to participate. More diagnoses per patient were formulated according to the DSM-5 Pediatric Interview than the CASDI, which is in agreement with the literature on fully structured diagnostic interviews [11].

In consistency with previous research [12-18], sensitivity of the CASDI was high (above 70%) for the great majority of the diagnostic categories. The low sensitivity rates (below 50%) of bipolar and related disorders have been also reported in other studies [16-18]. The lack of mechanisms to verify the veracity of the information regarding manic symptoms and the need of clinical judgment have been suggested as possible reasons for low sensitivity in diagnosing bipolar disorders through structured interviews [17]. Moreover, the CASDI showed low sensitivity in diagnosing dissociative disorders. The nature of dissociative symptoms and the limited capability of children and parents to provide this type of information based on Likert scale responses to structured questions may explain this low sensitivity rate.

It should be noted here that the CASDI required impairment of at least mild severity in order to diagnose a specific mental health disorder in children who attained symptom thresholds for this specific disorder. Although this requirement may decrease diagnostic thresholds, there is some evidence that a requirement of severe impairment criteria miss about half of the clinically referred cases of children who eventually remain undiagnosed and possibly untreated [19].

Standardization of data collection, widening and deepening the investigation field of the interviewer, standardization of the communication between different professionals, and reducing clinician bias are some of the advantages found in the CASDI as well as previously reported fully structured diagnostic interviews. Moreover, the additive value of the CASDI compared with other structured interviews is the integration of dimensional model of mental illness into the long-standing categorical model of psychiatric diagnostic interviews. By quantifying and personalizing the diagnostic criteria for a particular child and his or her caregiver according to their own subjective experience of core symptoms and the clinical judgment, when needed, and by employing computerized decision trees (algorithms), the CASDI seem to offer a convenient, easy to use and developmentally acceptable diagnostic procedure yielding a valid diagnosis.

However, the rigidity of the CASDI rules and the inflexibility of the diagnostic algorithms hamper the detection of symptom hierarchical importance that can only be detected in unstructured interviews that are mostly based on clinical judgement. Another possible limitation is that the interviewer is obliged to rely on the respondent’s assessment of whether a symptom is present or not, and each respondent may be using a scale of symptom severity that is different from that employed in clinical practice. For instance, it has been shown that children tend to overestimate the occurrence of rare phenomena that lie outside the realm of normal experience, such as manic or psychotic symptoms [20]. The possible child and adolescent interviewees' difficulty of understanding some items, and the lack of mechanisms to exclude misunderstanding and intentional or unintentional denial of symptoms as well as to verify the accuracy of recall of symptoms are also some of the CASDI limitations that the joint interview of both children and parents attempt to address to some extent. Finally, the need for some knowledge or clinical judgment capacity to identify some mental symptoms does not permit the administration of the CASDI by totally lay interviewers, even after strong training in the interview administration.

Overall, the CASDI has high validity (above 70%) in diagnosing the great majority of child and adolescent mental health disorders in Greek population. However, we should be cautious when using it for bipolar disorders and dissociative disorders. The observed validity indicators suggest that the CASDI can potentially be used to accurately diagnose child and adolescent mental health disorders for research and clinical purposes. Its use in clinical samples and special populations, such as hospitalized or institutionalized children and adolescents, deserves further study.