The Dark Side of the Moon: Meta-analytical Impact of Recruitment Strategies on Risk Enrichment in the Clinical High Risk State for Psychosis

Abstract

Background: The individual risk of developing psychosis after being tested for clinical high-risk (CHR) criteria (posttest risk of psychosis) depends on the underlying risk of the disease of the population from which the person is selected (pretest risk of psychosis), and thus on recruitment strategies. Yet, the impact of recruitment strategies on pretest risk of psychosis is unknown.

Methods: Meta-analysis of the pretest risk of psychosis in help-seeking patients selected to undergo CHR assessment: total transitions to psychosis over the pool of patients assessed for potential risk and deemed at risk (CHR+) or not at risk (CHR-). Recruitment strategies (number of outreach activities per study, main target of outreach campaign, and proportion of self-referrals) were the moderators examined in meta-regressions.

Results: 11 independent studies met the inclusion criteria, for a total of 2519 (CHR+: n = 1359; CHR-: n = 1160) help-seeking patients undergoing CHR assessment (mean follow-up: 38 months). The overall meta-analytical pretest risk for psychosis in help-seeking patients was 15%, with high heterogeneity (95% CI: 9%-24%, I (2) = 96, P < .001). Recruitment strategies were heterogeneous and opportunistic. Heterogeneity was largely explained by intensive (n = 11, β = -.166, Q = 9.441, P = .002) outreach campaigns primarily targeting the general public (n = 11, β = -1.15, Q = 21.35, P < .001) along with higher proportions of self-referrals (n = 10, β = -.029, Q = 4.262, P = .039), which diluted pretest risk for psychosis in patients undergoing CHR assessment.

Conclusions: There is meta-analytical evidence for overall risk enrichment (pretest risk for psychosis at 38 monhts = 15%) in help-seeking samples selected for CHR assessment as compared to the general population (pretest risk of psychosis at 38 monhts=0.1%). Intensive outreach campaigns predominantly targeting the general population and a higher proportion of self-referrals diluted the pretest risk for psychosis.

Keywords: CAARMS; SIPS; meta-analysis/risk; prevention; psychosis; schiz-ophrenia.

Fig. 1.

Pretest and posttest probability in prognostic tests.

Fig. 2.

Illustration of hypothetical recruitment processes and, relatedly, risk enrichment of pretest risk of psychosis in clinical high-risk (CHR) samples, adapted from. In the general population, mental problems, mainly anxiousness, depressiveness and family/partner problems, and/or worries about one’s mental state may lead to help-seeking. Thereby, outreach campaigns targeting the general population might lead to seeking help directly at high-risk services (self-referrals) but might also unspecifically increase symptom awareness and, relatedly, worries about one’s mental state and help-seeking, thus potentially resulting in a risk dilution. Outreach campaigns targeting (mental) health providers may lead to selective referrals from these sources suspected to possibly suffer from a CHR state; thereby pretest risk enrichment will likely be highest in referrals from mental health professionals.

Fig. 3.

Type of Outreach LYRIKS activities adopted by each of the CHR studies included in the current database. Outreach activities were defined and clustered as indicated in the community engaged framework adopted by the LYRIKS CHR study. These included the following domains: (A) screening assessments and recruitment, in green (outpatient and satellite clinics, armed forces, private hospitals, government organizations, internet gaming shops, and youth hubs); (B) workshops involving various community partners such as counsellors and mental healthcare professionals, in orange; (C) roadshows, in blue; (D) student internships, in gray; (E) print media, in violet (brochures and posters, articles and advertorials, newsletter); (F) social media, in yellow (facebook, twitter, blog, website). The LYRIKS domains were further expanded with the addition of (G) prescreening questionnaires, in red. MHP, Mental health professionals.

Fig. 4.

Meta-analytical estimate of pretest risk of developing psychosis (at an average follow-up of 38 months) in 2519 help-seekers at high-risk services reported in the 11 studies included. There was large and significant between-studies heterogeneity.

Fig. 5.

Metaregression of the number of outreach LYRIKS activities (see figure 3) adopted by each study and pretest psychosis risk (at an average follow-up of 38 months) in subjects recruited for CHR assessment.