Article Text
PDF
Statistics from Altmetric.com
Tyrone D Cannon
Correspondence to: Tyrone D Cannon, PhD, Department of Psychology, University of California, Los Angeles, 1285 Franz Hall, Los Angeles, CA 90095, USA; cannon{at}psych.ucla.edu
QUESTION
Question:
How common is conversion to psychosis in high risk youths, what factors predict this risk?
Population:
370 people (mean age 18 years) with prodromal psychotic symptoms (according to Structured Interview for Prodromal Syndromes (SIPS) criteria) who had been referred by school counsellors or community-based mental health professionals. The main SIPS criteria include: onset or worsening in the preceding 12 months of attenuated positive thoughts in one or more of five categories (grandiosity, unusual thought, paranoia/suspicion, disorganised communication, or perceptual abnormalities; rated from 0 (none) to 6 (fully psychotic state), score of 3–5 indicates a prodromal state); onset in previous 3 months of brief intermittent psychotic symptoms not meeting DSM-IV threshold for psychotic disorder; or the combination of deterioration of ⩾30% on the General Assessment of Functioning Scale in the past 12 months plus a genetic risk factor (either schizotypal personality disorder or a first degree relative with psychotic disorder).
Setting:
Eight clinical research centres in North America; time period not stated.
Prognostic factors:
77 possible prognostic factors in 10 domains were tested, including: sociodemographic factors, genetic risk factors, prodromal symptoms, social and role functioning, substance abuse, comorbid psychiatric diagnoses, and antipsychotic use during follow-up. Factors associated with progression to full psychosis in univariate analyses were tested in multivariate analyses within their domains, followed by multivariate analysis across the domains for the significant factors. The factors’ positive predictive power, sensitivity and specificity identified were examined for each factor individually and in all possible combinations.
Outcomes:
Time to full psychosis (based on SIPS criteria).
METHODS
Design:
Prospective cohort study.
Follow-up period:
30 months.
MAIN RESULTS
Seventy nine participants (21.4%) were lost to follow-up and were not included in the analyses. Of the remaining 291 participants, 82 progressed to full psychosis during follow-up (cumulative prevalence at 30 months: 35.3%). Across domains, only five factors independently predicted progression to full psychosis: social impairment (p<0.001), genetic risk factor plus functional deterioration in the past year (p = 0.001), paranoia/suspicion (p = 0.002), history of any substance abuse (p = 0.009), and unusual thought content (p = 0.01). The presence of each factor was similarly likely to correctly identify a person who would go on to develop full psychosis (positive predictive value: 43% to 52%), but the sensitivity and specificity were variable (sensitivity varied from 29% (substance abuse) to 80% (social impairment); specificity from 37% (paranoia/suspicion) to 83% (substance abuse)). Combinations of factors improved positive predictive value but reduced sensitivity. A model combining the presence of genetic risk factors plus functional deterioration, unusual thought content, and impaired social functioning gave the highest positive predictive value (81%; sensitivity 30%, specificity 90%).
CONCLUSIONS
Five independent prognostic factors have been identified that predict progression to psychosis in high risk youths. Prediction algorithms based on these factors could be utilised for recruiting similar youths into prevention programmes.
ABSTRACTED FROM
Cannon TD, Cadenhead K, Cornblatt B, et al. Prediction of psychosis in youth at high clinical risk. Arch Gen Psychiatry 2008;65:28–37.
Commentary
Schizophrenia has a prodromal phase that precedes the onset of psychotic symptoms, often by several years.1 A main aim of identification of the prodrome is to intervene early to prevent or at least delay onset of disorder.2 The challenge in detecting such a stage prospectively is to define a group with high likelihood of progression from non-psychotic state to psychotic disorder, while minimising labelling and unnecessary treatment of those who are not at risk.
An empirical definition of the “prospective prodrome” or “Ultra High Risk” (UHR) mental state was developed in Australia in the 1990s. These criteria were shown to confer a high risk of psychotic disorder within a short time frame.2 The criteria were adapted by US researchers. This paper reports on the findings of the North American Prodromal Longitudinal Study (NAPLS). With eight sites and 291 subjects, NAPLS is the largest cohort of “prodromal” subjects published to date.
Thirty five per cent of the subjects developed psychotic disorder within 2.5 years, with the main period of risk in the first year. This confirms the predictive validity of the criteria within a help-seeking population. It has clinical implications for timing of monitoring and preventive treatment. The list of predictors and their combinations are helpful for those working within, or setting up, UHR services. The methods of increasing positive predictive power (PPP) by narrowing the sample could be applied in these settings. However, a consequence of focusing only on an enriched group is the reduction in sensitivity, thus many young people who converted to psychosis would be missed. This may not be a problem if one aims, for example, to trial anti-psychotic medication in this group—the sacrifice of sensitivity to minimise false positives may be deemed worthwhile. However, if the goal is to maximise service coverage for those at risk, then narrowing the focus would not represent the best approach. Overall, the study findings may be helpful for those working with “prodromal” patients. However, the criteria should not be applied to the general population, in which their predictive validity would be much lower.
Supplementary materials
web only notes 11/3/72
The authors note that the algorithms generated by this study are not expected to be useful for screening the general population or people who are completely unaffected by psychotic symptoms. Further validation of the predictive value of the algorithms in other populations will be needed.
Footnotes
Source of funding: National Institute for Mental Health and the Staglin Music Festival for Mental Health.
▸ Additional notes are published online only at http://ebmh.bmj.com/content/vol11/issue3
Footnotes
Competing interests: AY received an Investigator Initiated grant from the Janssen Research Foundation for a treatment trial in the UHR group.