logo SBA


Digital archive of theses discussed at the University of Pisa


Thesis etd-07202017-223053

Thesis type
Tesi di specializzazione (4 anni)
Thesis title
Predicting long-term diagnostic instability of acute and transient psychotic disorders: an 8-year retrospective study based on South London and Maudsley electronic health records
Course of study
relatore Prof.ssa Dell'Osso, Liliana
  • electronic health records
  • diagnostic instability
  • ATPD
  • schizophrenia
Graduation session start date
Background: Brief psychotic episodes of dramatic symptomatology but remitting course represent one of the most intriguing paradoxes in psychiatry. They have received competing only partially overlapping operationalizations: acute and transient psychotic disorder [ATPD] in the International Statistical Classification of Diseases, 10th revision (ICD-10), brief psychotic disorder [BPD] in the Diagnostic and statistical manual of mental disorders (DSM-5), brief intermittent psychotic symptoms [BIPS], and brief limited intermittent psychotic symptoms [BLIPS] under the at-risk mental state paradigm. These constructs have high long-term diagnostic instability, mainly shifting to schizophrenia-spectrum disorders. However, the current state of knowledge does not allow the early prediction of diagnostic change to any non-organic psychotic disorders, and particularly to schizophrenia-spectrum disorders, in patients with a first index diagnosis of ATPD.
Methods: Electronic health record-based cohort study. Individuals were drawn from electronic, real-world, real-time health records relating to routine mental health care in South London and Maudsley (SLaM) National Health Service Trust in London, United Kingdom. The study included all patients who received a first index diagnosis of ATPD (F23, ICD-10) within SLaM between 1st April 2006 and 15th June 2017. We estimated the cumulative risk of diagnostic change from ATPD to any ICD-10 diagnoses of non-organic psychotic disorders, and specifically to schizophrenia-spectrum disorders with Kaplan-Meier failure function. A Cox proportional hazards model was used to evaluate the effect of socio-demographic and clinical predictors on the probability of diagnostic change to schizophrenia-spectrum disorders and to derive a prognostic index.
Results: A total of 3074 patients receiving a first index diagnosis of ATPD (F23, ICD-10) within SLaM were included in the study. The mean follow-up was 2200 days. The mean age was of 34 years, 53% were males, and 46% were of black ethnicity. The most frequent (70%) ATPD subtypes were F23.8 Other acute and transient psychotic disorders, and F23.9 Acute and transient psychotic disorders unspecified. ATPD recurred in 19% of our sample only. Approximately one third of episodes of ATPD lasted more than 30 days. An acute stress preceding ATPD was rarely recorded (1%). The cumulative 8-year risk of diagnostic change was: 1) 46.25% (95% CI 44.17%-48.37%) to any non-organic psychotic disorder; 2) 36.14% (95% CI 34.09%-38.27%) to schizophrenia-spectrum disorders. Gender, age*gender interaction, ethnicity, ATPD subtype, and non-rapidly remitting psychotic symptoms were associated with diagnostic change to schizophrenia-spectrum disorders. The predictive model based on these factors was used to stratify patients with a first index diagnosis of ATPD into 2 risk classes: low 29.67% (95% CI 27.22%-32.28%), and high 49.89% (95% CI 46.23%-53.68%).
Conclusions: Electronic health record-based data suggest that approximately half of patients with a first index diagnosis of ATPD will develop a long-lasting non-organic psychotic disorder over 8 years. In particular, 36% of patients with a first ATPD-presentation will shift to a disorder in the schizophrenia spectrum. Our prediction model based on socio-demographic and clinical features allows the stratification of patients at differential risk of pejorative outcomes. ATPD as a whole represents a highly unstable diagnosis, with substantial risk of transition to persistent psychotic disorders with poor prognosis and global functioning.