Screening of the college students at clinical high risk for psychosis in China: a multicenter epidemiological study

Since being proposed by Sullivan in 1927, the notion of early detection and preventive intervention of psychosis syndromes has received increased attention in the recent two decades [1‐4]. As a series of studies of first-episode psychosis suggests that longer duration of untreated psychosis significantly correlates with worse treatment response and prognosis [5], global researchers and clinicians have reached a consensus that early detection and prevention is central to contain the development and burden of psychosis. They started to investigate whether there is an earlier period in the development of psychosis that can be detected before the break, and then gave birth to the concept of ultra-high risk (UHR) or clinical high risk (CHR) for psychosis [6, 7]. CHR subjects refer to individuals who are identified as having nonspecific symptoms (e.g., cognitive symptoms) and attenuated or intermittent psychotic symptoms, schizotypal personality traits or having asymptomatic genetic risks, but do not meet criteria for a syndromal psychosis, and thereby they are considered to be at increased risk for developing psychotic disorders [4, 8]. Fusar-Poli et al. [9] reported that 20–35% CHR subjects developed syndromal psychotic disorder in 2 years through a meta-analysis, and Zhang et al. [10] discovered that the transition rate in the first 2 years was 26.4% in a Chinese help-seeking sample, both of which indicated that the risk of developing psychosis in CHR subjects deserved serious attention.

To diagnose and assess symptoms of CHR, two categories of instruments based on two complementary criteria have been globally recognized: the UHR criteria that focus on attenuated or brief limited intermittent psychotic symptoms and genetic risk, and the basic symptoms criteria that focus on cognitive and perceptive symptoms. The UHR criteria were designed to detect the risk for developing a first-episode psychosis in 12 months [11]. The most recognized instruments developed from the UHR criteria include the Structured Interview for Psychosis-risk Syndromes (SIPS) [12] and the Comprehensive Assessment of At-Risk Mental States (CAARMS) [13]. While the basic symptoms criteria aimed at detecting the risk for psychosis ideally before the appearance of functional impairments [11]. The most representative instruments are the Schizophrenia Proneness Instrument, Adult (SPI-A) [14] and Child & Youth version (SPI-CY) [15]. Since the instruments listed above are based on the clinician’s assessments through clinical interviews, several self-report scales have been established to simplify the screening of CHR individuals, such as Prodromal Questionnaire-Brief version (PQ-B) [16], Prodromal Questionnaire-16 items (PQ-16) [17], Prime Screen [18] and Early Psychosis Screener (EPS) [19, 20].

Owing to the development of the diagnostic instruments, the screening of CHR have been conducted globally, especially in samples of clinician-referred patients and help-seeking individuals. However, since some of the CHR subjects won’t see a doctor (e.g. because of social discrimination, stigma or poor knowledge of psychosis) before the onset of psychosis or significant functional impairments [21], screening CHR in clinical samples appeared to be a passive method. Therefore, detecting CHR in general populations have raised more and more concerns recently. Besides, as the risk age range of CHR is 15 to 22 years, great attention was paid to the studies carried out in students in universities and high schools [22]. Notwithstanding, detection of CHR in China is still limited by its passivity that the clinical interviews and rating scales can almost only be approached in psychiatric clinics or wards, which indicates that only individuals with help-seeking behaviors can be tested, and doctors should always wait passively for the suspected cases. In fact, most researches concerning CHR detection in China were conducted in help-seeking individuals [10, 23, 24]. Only few studies attempted to detect CHR in Chinese general populations. For instance, Chen et al. established a two-stage screening method (PQ-16 followed by SIPS) in 579 Chinese university students who were enrolled in a mental health education course, and reported that the prevalence of CHR was 1.1% [25]. However, the prevalence rates of CHR varied widely in the literatures, ranging from 0.3 to 2.4% [26‐28]. Therefore, an epidemiological study of CHR detection conducted in a larger sample of Chinese generalpopulations with established psychosis-risk instruments will be of great significance. In this study, we generalized a 3-stage method (consists of PQ-B, telephone interview and SIPS) for the detection of CHR individuals in college students from Shanghai, Nanjing and Nanchang in China. The present study aimed to investigate the screening procedure, and explore the characteristics of prodromal symptoms of psychosis in Chinese college students. To our knowledge, this is the first multicentre epidemiological study on the detection of CHR subjects in Chinese college students.

The detection of CHR samples in both the clinical settings [10, 33‐36] and non-clinical settings [26‐28] attracted a great deal of attention around the world, while the detection rate of CHR subjects in these studies varied considerably. Previous studies reported that the prevalence of CHR in the clinician-referred samples was 32–42% [33‐35], which correspond roughly to one another. While the detection rate in samples of help-seeking individuals ranged from 4.2 to 80%, which varied from site to site [10, 37‐39]. For the general population (primarily focused on young adults and adolescents), the annual incidence of new cases of CHR was estimated at 1/10000 [33], and the prevalence ranged from 0.3 to 2.4% [26‐28]. The diagnostic instruments were considered to be a great factor contributing to the inconsistency of the detection rate. Schultze et al. [27] screened CHR in a young adult community with SIPS and SPI-A, and reported that 2.4% of the participants met at least one CHR criterion, while only 0.6% met the APS criteria identified by the SIPS. Salazar et al. [40] reviewed 56 studies and suggested that the 0.3% of the general non-help-seeking population of youth met the DSM-5-APS criteria, while 1.3% of the general population met the SIPS-APSS criteria.

In this study, the detection rate of CHR subjects identified based on the SIPS criteria in Chinese college students was 0.3%. Although it was lower than the detection rate (1.1%) reported in another study conducted in the Chinese college students [28], a multicentre sample and a random sampling method both proved the reliability of the results in a degree. Nevertheless, as is mentioned before, only about 30% of the CHR would make the transition to psychosis, and thus the prevalence of CHR should exceed that of psychotic disorders intheory. While in fact, the prevalence of schizophrenia and other psychotic disorders was about 0.7% [32], much higher than the detection rate of CHR reported in this study. A possible explanation for the contradiction may originate from the education level of the sample. Since only college students were recruited in this study, a large proportion of the CHR individuals who failed or even quitted the College Entrance Exam because of the disturbance of prodromal symptoms were not investigated, and thus it may be underestimated if regarding 0.3% as the prevalence of CHR in the Chinese general population.

Since first-episode psychosis usually occurs in late adolescence and early adulthood, CHR or prodromal psychosis were believed to have an earlier onset [41, 42]. Our research demonstrated that the peak age range of the CHR identification by SIPS was between 17 to 20 years in the sample of Chinese college students, which was largely in line with previous studies, but somewhat narrower (e.g., SHARP reported 16 to 21 years in a Chinese sample) [10, 22]. The mean age of CHR was 18.3 (SD = 1.5) years in this sample, which was similar to the sample of NAPLS [33], but younger than that of SHARP [10], and older than that of RAP [22]. The demographic characteristics of the participants enrolled in these studies may account for the age discrepancy. As the participants recruited in this study were college students with an age range of 14 to 31 years, while NAPLS recruited clinician-referred subjects aged 12 to 35 years, SHARP included help-seeking subjects aged 15 to 45 years, and RAP included help-seeking subjects aged 12 to 22 years.

Similar to previous studies using PQ-B to screen CHR in Chinese population [16, 30, 43‐45], our research proved that PQ-B could be a feasible first-step tool to detect suspected CHR subjects in college students. Both the sensitivity and specificity of PQ-B were excellent in this study. Also, the results showed that 8 items of PQ-B including the PQ1 (asking “Do familiar surroundings sometimes seem strange, confusing, threatening or unreal to you?”), the PQ3 (asking “Do things that you see appear different from the way they usually do (brighter or duller, larger or smaller, or changed in some other way)?”), the PQ4 (asking “Have you had experiences with telepathy, psychic forces, or fortune telling?”), the PQ12 (asking “Do you worry at times that something may be wrong with your mind?”), the PQ13 (asking “Have you ever felt that you don’t exist, the world does not exist, or that you are dead?”), the PQ15 (asking “Do you hold beliefs that other people would find unusual or bizarre?”), the PQ17 (asking “Are your thoughts sometimes so strong that you can almost hear them?”) and the PQ18 (asking “Do you find yourself feeling mistrustful or suspicious of other people?”) had significant predictive value of SIPS diagnosis, which indicated that delusional ideas, perceptual abnormalities and suspiciousness were key symptoms in identifying CHR subjects. Similarly, Xu et al. [30] and Schultze et al. [27] also reported that perceptual abnormalities, delusional ideas and suspiciousness were the most prevalent psychosis-risk symptoms.

Nevertheless, our results recommended higher cutoff points for the total score and distress score of PQ-B (13 and 46) than those of SHARP did in the Chinese population (7 and 24, 30], which were also much higher than the cutoff points reported in the initial validation study of PQ-B (3 and 6, 16]. Despite the cultural background and language may lead to the differences of PQ-B cutoff points [46], there are still other reasons underlying the differences between the present study and SHARP. First, the participants in this study were college students who were tested with PQ-B in non-clinical settings (online or in the colleges), while SHARP recruited help-seeking outpatients from a psychological counselling center [30]. As prodromal symptoms mentioned in PQ-B were easily mixed with psychotic-like experiences (PLEs) that commonly existed in the general population [47‐50], subjects who only had PLEs may also be tested positive with PQ-B, causing false positives. In addition, being rated in a clinical setting, rather than a non-clinical setting, may help the participants better understand the prodromal symptoms mentioned in the scale, and thus reducing false positive errors. Indeed, if we used the same cutoff points as SHARP did in their screening procedure, 72.0% (16,595/23063) of the participants would be screened positive with PQ-B in this study, while the positive rate of PQ-B in SHARP was 62.1% (1681/2705, 10], which may indicate that our false positive rate in PQ-B screening was higher. Second, the screening procedure differed. SHARP employed a 2-stage method where the PQ-B-positive subjects would all be invited to the SIPS interview, and 18.3% of the PQ-B-negative subjects were also interviewed [10]. However, we used a 3-stage method that applied a telephone interview to the PQ-B-positive subjects, and only subjects screened as positive in the telephone interview were invited to the SIPS interview. Moreover, no further investigations were provided to the negative subjects in the present study. These variance in the screening procedure may have contributed to the underestimation of CHR subjects, which would interfere the examination of the psychometric properties of PQ-B.

As discussed in the last paragraph, the telephone assessment can cause the missed diagnosis of CHR, which was a limitation of this study, whereas, it saved a lot of human resources in the process of CHR screening, in fact, particularly in a large sample of general population. It is worth mentioning that the telephone assessment employed in this study was not a semi-structured or structured interview that can be seen in other studies, although we also use SIPS as the assessing instrument [27, 51, 52]. In the present study, researchers only needed to ask questions following the SIPS subscale for positive symptoms, and to rate the severity for each domain based on the participants’ subjective reports. Consequently, it only took 5 min on average, which was much time-saving compared to a structured/semi-structured interview. In fact, it has also enhanced the screening efficiency by preventing almost 90% of the PQ-B-positive subjects from receiving a face-to-face SIPS interview, despite the fact that it was, of course, based on the sacrifice of the detection rate.

Besides, another limitation also lies in the screening process of GRDS, a subtype of the CHR syndromes. It’s worth noting that only participants who received the SIPS interview were screened for GRDS in this study. In other words, those participants who had family history of psychosis but did not manifest attenuate psychotic symptoms were neglected for the possible diagnosis of GRDS. Because for those first degree relative with psychosisor schizotypal personalitydisorder traits, a 30% drop in GAF score, rather than the attenuate psychosis symptoms is a necessity when diagnosing GRDS [12]. Therefore, the prevalence of CHR was underestimated in the present study.

Lastly, due to limited resources, the information received from the participants was incomplete, which is also a limitation of this study. On the one hand, we missed some important clinical features of the subjects, including family history and psychiatric comorbidity. On the other hand, the CHR subjects were not followed after being diagnosed, and therefore the transition to psychosis or other mental disorders was not investigated in this study.

In conclusion, the 3-stage method is a useful strategy for screening CHR individuals in the general population, especially when the sample size is large. Among the college students, those with delusional ideas, perceptual abnormalities or suspiciousness were more likely to be screened as CHR subjects. However, several modifications should be made to the 3-stage method, especially for the screening of the familial high risk individuals, in order to enhance the sensitivity of the screening process, and further investigations comparing the 3-stage method with others can be done to explore more efficient ways of the screening. Also, for future researches, a longitudinal follow-up of the CHR subjects is essential to investigate the transition to psychosis of the CHR subjects, and to explore the related risk factors.