The Programma2000, operating in Milan since 1999 under the Health Authority of the Niguarda Ca’ Granda Hospital, was the first center to be opened in Italy with the aim of providing early detection and intervention on people with psychosis at its onset [16, 17], and served as a pilot program for the development of other centers in the area.

The Programma2000 covers a catchment area catering for approximately 200,000 inhabitants.

The early intervention center of the Civic Hospital of Desio and Vimercate was established in 2006 and covers a catchment area of about 230,000 residents, north of Milan.

The early intervention center of the Civic Hospital of Legnano (Milan) was established in 2005 and covers a catchment area of about 370,000 residents.

The early intervention center of the Psychiatric Unit of Bollate of to the “G. Salvini” Hospital (Garbagnate Milanese) was established in 2006 and covers a catchment area of about 95,000 residents.

The four catchment areas together cater for 895,000 inhabitants.

Each center is lead by a senior psychiatrist and its staff includes: two or more psychiatrists, operating in the center on a rotational scheme; two or more licensed psychotherapists; two or more clinical psychologists; two or more educators or nurses; a variable number of psychologists and educators in training.

All centers provide prompt intervention (within 24 hours) to the referred patients, who are offered a comprehensive, multidimensional evaluation with a package of standardized assessment instruments aimed at evaluating general psychopathology, level of functioning and associated impairment, disability and cognitive deficits [details in 16, 17].

All centers operate on an outpatient basis; admissions, when necessary, are negotiated with the General Hospital Psychiatric Units (GHPUs) operating in the same area. The beds available for hospital treatment were 10 to 15 per 100,000 when the study was performed.

All interventions are free of charge for the enrolled patients and are financed through a special grant from the Lombardy Regional Authority. Further details on the organization of the mental health system operating in Lombardy were reported elsewhere [18].

The patients were assessed using the following standardized assessment instruments: (i) a socio-demographic form; (ii) the Early Recognition Inventory Retrospective Assessment of Symptoms checklist (ERIraos-CL), a 17-item screening checklist intended to select persons needing a more in-depth assessment [19, 20]; (iii) the Health of the Nation Outcome Scale (HoNOS), to assess psychopathology and disability: it includes 12 five-point items to evaluate clinical and social functioning over the prior 2 weeks [21, 22]; (iv) the 24-item Brief Psychiatric Rating Scale (BPRS), to assess general psychopathology [23, 24]; (v) the Global Assessment of Functioning (GAF) [25]; (vi) and (vii) the World Health Organization Disability Assessment Schedule (WHODAS) version II [26].

Patients were included in the study if they were between 17 and 30 years of age, and had been referred to any of the four participating centers after a first contact with any public mental health service of the catchment area for a first episode of psychosis (i.e. they had never received antipsychotic treatment before the current episode). Referral sources were mental-health professionals and associated surgeries, family physicians, or direct family referrals in response to awareness campaigns; self-referral was also allowed.

The main criterion for inclusion was a diagnosis of schizophrenia or related syndromes (F20-29 in ICD-10) according to the ICD-10 criteria [27]. Affective psychosis (bipolar disorder, or unipolar disorder with psychotic features) was an exclusion criterion, as was a co-morbid persistent substance-use dependent disorder, while substance use/abuse without dependence was not. Data on substance use, abuse and dependence were based on a detailed interview with the patient and a key informant (a close relative, usually a parent). Substance abuse was investigated by asking whether the patient had received any diagnosis of substance abuse in the last 24 months, or habitually consumed a drug from a list including alcohol, tobacco, cannabis, cocaine, heroin/opiates, amphetamine and its derivatives, hallucinogens and a residual class of “others”. The threshold for abuse was defined according to the ICD-10 criteria, i.e. the inability to stop using the drug despite awareness of a health risk [27].

The past diagnoses received by the patients in the 12 months preceding the enrollment were considered too. In Lombardy both public and private psychiatrists are expected to provide a diagnosis according to ICD-10 criteria to justify the prescription of a drug treatment [18].

Additional information was collected during the interview of the patient and the key informant. Duration of untreated illness (DUI) and DUP were both measured as the time elapsed from the onset of key symptoms (anxiety, depression or social withdrawal for DUI; hallucinations, delusions or bizarre behavior for DUP) to the beginning of treatment (pharmacotherapy or psychotherapy) prescribed by a psychiatrist; DUP was measured in days, DUI in months. To measure DUP/DUI, researchers considered the symptoms as they were elicited by the ERIraos-CL, and the patient’s estimated time of onset of key symptoms as listed in the tool. A therapist (usually a psychiatrist) and a researcher (usually a psychologist or an educator) of the team made the DUI/DUP assessment jointly. In problematic cases consensus with a senior clinician was sought.

The enrolled patients received a comprehensive, tailored and flexible intervention package. Prescribed tailored interventions included individual psychoeducational and motivational sessions, cognitive-behavioral psychotherapy, individual family psychoeducation and support, therapeutic group activities (e.g., anxiety management, assertive and problem-solving training, substance abuse prevention, etc.), various social group activities (e.g., music, multimedia, empowerment, computer training sessions, language classes, etc.), and supportive interventions on employment, school, compliance with medication, and planning of recreational activities [16, 17].

The institutional review board of the participating centers and of the INIH approved the protocol of the study, which conforms to the provisions of the 1995 Declaration of Helsinki (as revised in Tokyo, 2004). All patients gave their informed consent.

All analyses were carried out with the Statistical Package for Social Science version 17.0. All analyses were two-tailed, and statistical threshold was set at p < 0.05.

Cluster analysis was used to group subjects into clusters identified by typological characteristics that were not defined a-priori, were homogeneous within each cluster and heterogeneous between clusters. A two-step method was used to deal with both continuous and categorical variables and to determine the optimal number of clusters automatically. Likelihood ratio was used to determine the minimum distance between clusters across variables, in order to minimize within-cluster variation and maximize between-cluster variation. Model selection of grouping was based on the Bayesian Information Criterion (BIC) [28]. Chi-square for categorical variables, and the Student t-test for continuous variables were used to compare clusters on the variables that defined the model.

Multi-dimensional scales were entered in the cluster analysis after their reduction to the main factors. Since there is no consistent factor structure for both the BPRS and the HoNOS [29-32], principal component analysis with Varimax rotation was carried out de novo on both the BPRS and the HoNOS items (separately). The number of factors to be extracted was determined according to the scree-plot method [33]. Factor scores were calculated by summing up the score values of the items that saturated the factor. Six factors were extracted for the BPRS that explain 64% of total variance; and four factors were extracted for the HoNOS, explaining 60% of total variance (see Table 1 for details; solution available upon demand).

The following variables were entered to define the clusters: sex; age; education; severe agitation in current episode; severe depression or apathy in current episode; severe distress, anxiety or phobia in current episode; alcohol abuse in current episode; substance use in current episode; severe poor self-care in current episode; functional, social or work impairment in current episode; severe social withdrawal in current episode; HoNOS Behavioral problems, and HoNOS Mood and physical disability problems. All the other variables listed in Table 1 were used to further characterize the clusters once they had been defined.

Patients were reassessed after six months of treatment. There were no dropouts in the first six months of treatment.

Patients improved on almost all symptomatic dimensions on the HoNOS and the BPRS, with greater improvement in cluster 1 than in cluster 2 (Table 2).

The differences were more evident for the HoNOS factors concerning “Cognitive and social functioning problems” and “Mood and physical disability problems”. However, the

greatest difference between clusters concerned the BPRS factor “Negative orientation towards others”, which decreased in cluster 1 but remained unchanged in cluster 2.

Some limitations of the study must be acknowledged. This study uses a large array of variables to define patient status. As a matter of fact, with so many variables sample size might have prevented the definition of more fine-grounded differences among clusters. Follow-up was short, too, and this might have limited inference on the long-term outcomes of the two identified clusters of FEP patients. Nevertheless, the findings of the study point to the importance of taking into account the variability of symptoms presentation in FEP patients and its impact on short-term outcome.

The setting, beside the limited sample size and the short follow-up, has influence on the generalizability of the findings. The sample was enrolled in four early intervention centers that operate on an outpatient basis and exclude patients with co-morbid persistent substance-use dependent disorder. The findings can be generalized to a subgroup of all FEP patients accessing psychiatric services in Italy.

The findings of this study need replication in larger samples and with a more extended level of severity. The early detection and treatment of cases with psychosis in need of care is expected to reduce morbidity and its related disability, and to improve the long-term outcome [46, 47]. Indeed, early intervention protocols were found effective on the short term and able to sensibly reduce both direct and indirect costs of care as a reflection of their effectiveness [48-50]. The evidence on the medium-term effectiveness of these programs is unconvincing [51-53], and their cost-effectiveness has been questioned [54-56]. Therefore, further studies are necessary to determine whether “heterogeneity-sensitive” services are actually more effective than “generalist” services.

Whatever the real effectiveness of third-tier services dedicated to the early detection and treatment of psychosis, there remains the need of identifying different profiles of symptoms presentation in FEP patients, since these different profiles can impact on short- and long-term outcomes. Albeit the findings of this study are in need of replication, we think that the two clusters we identified must be distinguished in clinical settings. To this aim, a detailed assessment is necessary to gain insight of the profile of patient's symptoms with attention being paid to both positive and negative symptoms. Negative symptoms such as anhedonia, apathy, autism and avolition are often overlooked in the Italian mental health system of care, with greater attention being devoted to the diagnosis and treatment of the symptoms that are more disturbing to the community: agitation and the disordered behaviors triggered by delusions and hallucinations.

Policymakers should recognize the importance of assessment in the treatment of severe mental disorders and fund appropriate protocols to diffuse the practice of assessing clinical status at enrollment and applying state-of-the-art outcome measures to the follow-up.