Early signs and symptoms of psychosis among Palauan adolescents


Dr Paul Florsheim, Center for Addictions and Behavioral Health Research, University of Wisconsin-Milwaukee, 1020 N 12th Street, Suite 4180, Milwaukee, WI 53233, USA. Email: paulf@uwm.edu


Aim: This study was designed to identify early symptoms associated with the occurrence of psychosis during adolescence.

Method: Participants were recruited in the Republic of Palau, an isolated island nation in Micronesia with a prevalence rate for schizophrenia of 1.99%. Diagnostic interviews were used to obtain reports of early and current symptoms from 112 genetically high-risk (GHR) and 208 genetically low-risk (GLR) adolescents (ages 16–23). Based on current psychotic symptoms, participants were sorted into three groups: non-clinical, at-risk/symptomatic risk and clinically symptomatic.

Results: Multivariate analysis of variance revealed several between-group differences on rates of early symptoms. Most notably, youth who were in the GHR-clinically symptomatic group reported significantly higher rates of early marijuana use than GLR-clinically symptomatic youth, who were significantly more likely to report early symptoms of depression and behaviour disorders. In addition, several gender based differences in the link between early symptoms and adolescent onset psychosis were noted.

Conclusions: Findings are generally consistent with previous research on early indicators, though several unexpected findings suggest that results from this study may not be fully generalizable beyond this relatively isolated and culturally distinct Micronesian nation.

Identification of early signs of adolescent onset psychotic disorders could facilitate the development of strategies for early intervention, possibly delaying psychosis or diminishing its severity.1 Despite the development of criteria for identifying young people at heightened risk of developing a psychotic disorder,2 several difficulties and problems remain in implementing research on childhood symptoms of adolescent onset psychosis:

  • 1Because adolescent-onset psychosis has a low prevalence in the general population, prospective research designed to identify early symptoms requires extraordinarily large samples, followed over long periods of time. This is true even when considering the development of criteria purporting to identify young people at heightened risk of developing a psychotic disorder by groups such as the Personal Assessment and Crisis Evaluation (PACE) Clinic in Melbourne, Australia3 and researchers from Bonn, Germany.4 Although some capacity to accurately identify young people at heightened risk of psychosis in a prospective way has been reported by these groups,5–7 the reliability of these criteria used to identify these groups is questionable as evidenced by recent reports of falling rates of transition to full psychosis.8 Until the reliability of prospective predictors can be assured, the use of retrospective reports of youth who experience psychosis, whilst not ideal, remains necessary to facilitate the development of models for early detection.
  • 2Researchers studying the precursors of psychosis tend to focus only on young people who present at clinical settings with some signs of psychosis and who meet the criteria referred to above.9–11 As a result, conclusions drawn from these studies cannot be fully applied to individuals who do not meet those criteria. In other words, if an entire sample is thought to be at heightened risk for psychosis, then symptoms that have a high base rate in the general population (such as marijuana use, depression or anxiety) could be misconstrued as risk factors. Studies that include a comparison group that do not meet ‘at-risk’ criteria or have not experienced psychotic symptoms could help identify factors specifically associatedo with adolescent onset psychosis by differentiating between those symptoms that have a high base rate in the wider population of young people – and therefore could also be heightened in at-risk samples – and true, specific risk factors. Only five published studies have included a comparison group that does not meet heightened psychotic risk status. No conclusions about the specificity of risk factors for psychosis can be drawn from most of these studies7,12–15 because putative risk factors were not compared between the ‘at-risk’ cohorts and the comparison groups, or the comparison groups included young people who were experiencing some level of psychological difficulty rendering it difficult to make conclusions about the specificity of risk factors for psychosis.

Only one paper published to date has compared the experience of psychotic and other psychological symptoms between individuals who met prodromal psychosis criteria and a number of comparison groups (young people without any psychiatric history, young people with a first degree relative with psychosis and young people who met schizotypal personality disorder criteria).15 It was reported that young people identified as being at risk of developing a psychosis due to the experience of sub-threshold psychotic symptoms were clearly differentiated from the other groups across a range of demographic, symptom and functional domains assessed at entry to the study. Although this study was cross sectional in nature and did not incorporate information about transition to psychosis in either the ‘high-risk group’ or other participants, the authors concluded that these results were strong evidence for the diagnostic validity of the ‘prodromal risk syndrome’. Follow-up of the high-risk cohort was described in paper by Cannon et al. (2008).9 Five factors that were assessed at baseline were reported to be particularly important in the prediction of the onset of a psychotic disorder in the high-risk subjects: (i) family history of psychosis in conjunction with a recent deterioration of functioning; (ii) increased levels of unusual thought content; (iii) higher levels of suspicion or paranoid thinking; (iv) higher levels of social impairment; and (v) a history of substance abuse. As the comparison groups that were described in the paper by Woods et al.15 have not been followed-up, conclusions that can be drawn from the Cannon et al.9 paper regarding the specificity of any risk factors are limited. This suggests that there is a need for further studies that address this issue.

  • 3Despite recent advances in the epidemiology of genetic risk for psychosis,16,17 most researchers studying early onset psychosis have not targeted adolescents at increased genetic risk, although some have included family history of psychosis in their determination of risk status.11 Recruiting adolescents at genetic risk for psychosis provides a pragmatic approach to identifying early indicators of early onset psychosis. The Edinburgh High Risk Study has followed-up a cohort of young people who have a family history of psychotic illness and a healthy comparison group. They have compared the groups on a range of measures including psychopathology, neuropsychology and brain structure and function and have found differences between the high-risk group and the comparison group on a number of factors.18–22 The authors do note that the genetic component of schizophrenia affects many more individuals than will develop the illness, so research investigating risk factors for psychosis cannot only focus on individuals with a family history. Comparing the early symptomatic histories of adolescents at high and low genetic risk for psychosis could also help clarify important differences in the phenomenology of psychosis; the pattern of early symptoms may vary, depending on whether or not a clinically affected youth has a family history of psychosis.

In this study, our approach to identifying childhood predictors of adolescent onset psychosis was to: (i) include adolescents with psychotic symptoms and a non-psychotic comparison group; (ii) include adolescents with and without family histories of psychosis; (iii) gather life history reports of symptoms to determine which symptoms, recalled from earlier stages of life, were significantly associated with current reports of psychosis; and (iv) assess current level of psychotic symptoms in line with a simplified ‘clinical staging’ model, allowing researchers to move beyond a dichotomous definition of psychotic disorders.23 The aim of the study was to identify differences in early symptom profiles between symptomatic and asymptomatic adolescents at high and low genetic risk for psychotic disorders.

In line with previous research in high-risk populations as well as retrospective reports of earliest signs of psychosis, it was hypothesized that the development of psychosis in early adolescence would be associated with cannabis use during the early teenage years,24–26 mood and anxiety symptoms and the experience of sub-threshold psychotic symptoms – both hallucinations and delusions.27–29 We also hypothesized that the high genetic risk participants would report higher levels of psychopathology than the low genetic risk participants.


This study is based on a sub-sample of adolescent participants from The Palau Early Psychosis Study, which focused on Palauan adolescents, ages 13–23, at both high and low genetic risk for psychosis.13 Palau is an island nation in Micronesia with a prevalence rate of schizophrenia of 1.99% and a prevalence of psychotic disorders of 2.67%.30 Because Palau is home to a relatively isolated population of families at increased risk for schizophrenia and psychosis, it provides the unusual and unique opportunity to identify early signs of psychosis.

The Palau Early Psychosis Study was designed to explore clinical, psychosocial and neuro-cognitive functioning of adolescents who were either (i) at genetic risk because of their relationship to affected family members; (ii) at clinical risk because they reported some symptoms of early psychosis (but had no close affected relatives); or (iii) at low risk because they reported no significant symptoms of psychosis and had no familial history of psychotic illness. Young people who were recruited into the genetic high-risk group (GHR) were either offspring of a parent with a psychotic illness, predominantly schizophrenia, or offspring of an unaffected parent with two or more affected siblings. Clinically affected and non-affected participants in the low genetic risk group (GLR) were recruited from surveying high school students in four Palauan high schools. All GLR participants had no known family history of psychosis, based on parent report. Among the GHR, family history of psychotic disorder was confirmed or disconfirmed in a previous study by Myles-Worsley and colleagues (1999) using the Schedule for Affective Disorders and Schizophrenia – Lifetime Version [SADS-L31] or reviewing Belau National Hospital psychiatric records. Family pedigrees were developed for all participants to clarify genetic risk status.

In the present study, we included participants aged 16 years and older in order to help clearly differentiate between current symptoms and symptoms recalled occurring prior to the age of 14. We selected the age of 16 for inclusion because previous research has indicated that the risk of onset of schizophrenia and other severe mental illnesses increases around this age.32–34 We selected the age of 13 as the cut-off for early symptoms to help clearly differentiate between current and early symptoms, while allowing for the inclusion of the entire pre-pubescent phase of childhood. Additionally, childhood-onset schizophrenia, defined by an onset of psychotic symptoms before 13 years of age is a rare and severe form of the disorder that can be differentiated from adolescent- and adult-onset schizophrenia.35,36 Therefore, by only including participants aged 16 and above, we aimed to minimize confounding between two potentially different forms of the illness. The mean age of participants in the sub-sample was 17.44 years (SD = 1.33). None of the participants were in psychiatric treatment at the time of recruitment.


Data on specific symptoms across a range of psychiatric disorders was collected using a modified version of the Kiddie Schedule for Affective Disorders and Schizophrenia Interview – Present and Lifetime version (K-SADS-PL37). The K-SADS-PL was modified to (i) include prodromal and early psychotic symptoms as operationally defined by the Comprehensive Assessment of At Risk Mental States (CAARMS11); and (ii) consider cultural norms in Palau. Cultural modifications included careful inquiry regarding symptoms that might be considered delusions or hallucinations in a Western context, but could be associated with commonly held traditional beliefs in Palau. This was achieved by ensuring that all interviews were conducted by a native Palauan adult who is a community elder and has a well-grounded knowledge of cultural and traditional Palauan beliefs but who also has a nursing background and has received training in mental health assessment and particularly in the administration of the K-SADS. K-SADS-PL items that were not assessed were enuresis, encopresis, eating disorders and post traumatic stress disorder.

Current psychosis

Two scoring systems were used to assess the current level of psychotic symptoms.

First, two reliable research diagnosticians independently utilized standard DSM-IV guidelines to identify adolescents who met the criteria for schizophrenia, independently. Eighteen of the GHR group participants (16%) and 21 of the GLR (10%) met criteria for a psychotic disorder at the time of interview for this study. These 39 participants form the Transitioned to Psychosis cohort, referred to as TP in all subsequent discussion.

Second, a ‘psychosis severity’ scoring system for participants not in the TP group was developed, based on a summation of the severity and frequency of each psychotic symptom identified using the K-SADS (delusions, hallucinations, negative symptoms and other positive symptoms). Scores were calculated by summing the severity score, ranging from 0 (absent) to 6 (severe) and frequency score, ranging from 0 (absent) to 6 (continuous) for each symptom. Scores ranged from 0 (no symptoms) to 86 (moderately symptomatic). Using the CAARMS Unusual Thought Content, Non-Bizarre Ideas, Perceptual Abnormalities subscale and Disorganized Speech subscales to differentiate between low and moderate symptoms of psychosis, participants who were not in the clinically affected (TP) group were subdivided into two other groups of GHR and GLR participants. The GHR-NFS and the GLR-NFS groups included participants who reported ‘none or few symptoms’ of psychosis (NFS) and were determined to be at low clinical risk based on CAARMS criteria (score of 0 or 1). The GLR-MS and GHR-MS groups included participants were reported ‘moderate symptoms’ of psychosis (CAARMS score of 2–4), but determined to not have a clinical diagnosis.

Early symptoms

Psychiatric symptoms that were experienced prior to age 14 were assessed across the following domains of the K-SADS: depression/dysthymia, anxiety (anxiety disorders assessed include general anxiety disorder, panic disorder, separation anxiety disorder, obsessive compulsive disorder and phobias), attention-deficit/hyperactivity (ADHD), oppositional defiant and conduct disorder (ODD/CD), delusions, hallucinations, mania and marijuana use. Based on adolescent recall, early symptoms were quantified as the number of symptoms in each category of psychopathology recalled as occurring prior to the age of 14. As indicated earlier, we were primarily interested in identifying childhood indicators of adolescent onset psychosis and 13 is regarded as a rough marker for the end of childhood and the onset of adolescence in Palau (and elsewhere). Grouping early symptoms into a broadly defined time frame (before age 14), rather than relying on reports of specific age of onset, was intended to diminish error in retrospective reporting.38,39 Previous research has indicated that adolescents are able to reliably recall events in broad time frames, but that asking adolescents to identify specific time frames (such as a particular age) will increase reporting error.40

Data analysis

Based on the classification system described above, six groups were included in subsequent analyses: GLR-NFS, GLR-MS, GLR-TP, GHR-NFS, GHR-MS and GHR-TP. Preliminary analyses were run to examine group differences in age, gender and school status (enrolled/graduated vs. dropped-out). Chi-square analysis results indicated that more females were categorized in the GHR-TP group, although the difference was not statistically significant, χ2(2) = 5.55, P = 0.062. anova results indicated that participants in the GHR group were significantly older than participants in the GLR group (17.52 vs. 17.12), F(1319) = 8.19; P < 0.01, and females were significantly younger than males (17.15 vs. 17.45), F(1319) = 4.01; P = 0.046. Chi-square analysis also revealed no significant differences in the number of participants who were enrolled in or graduated from high school and those who had dropped out of school between genetic groups (enrolled/graduated GLR 95.7% vs. GHR 90.2%, χ2(1) = 2.872, P = 0.090 with Yates' correction) or between clinical status (enrolled/graduated NFS 94.4% vs. MS 93.1% vs. 94.7%, χ2(2) = 0.226, P = 0.893.) Age and gender were included in subsequent analyses as control variables.

Multivariate analysis of variance was run to examine between group differences in early symptom score. Gender, clinical status (NFS, MS and TP) and genetic status (GHR and GLR) were included as independent variables. Early symptom scores were included as the dependent variables.


Means and standard deviations of symptom scores across clinical, gender and genetic groups are presented in Table 1. Multivariate analysis results, using Wilks' Lambda, indicated a main effect for gender, F(8, 300) = 6.71, P < 0.01, genetic risk status, F(8, 300) = 5.01; P < 0.01, and clinical status, F(16, 600) = 13.27, P < 0.01. Results of univariate analyses outlined in Table 2 indicate that females were more likely than males to recall early symptoms of anxiety, mania and delusions. Males recalled more early symptoms of depression, conduct or oppositional disorder symptoms and higher rates of marijuana use than females. When including NFS, MS and TP participants, GHR participants were significantly more likely to recall early marijuana use than GLR participants. GLR participants were more likely to recall early ADHD symptoms and hallucinations. Regardless of genetic status, participants diagnosed with current psychosis (TP subgroup) were more likely to recall higher rates of early marijuana use, hallucinations, delusions, and to recall symptoms of depression, anxiety, ODD/CD and ADHD than all non-TP participants (MS and NFS subgroups). Participants in the MS group (again, regardless of genetic risk status) reported early delusions and hallucinations at a higher rate than participants in the NFS group.

Table 1.  Number of early symptoms by gender, clinical status and genetic risk status
Symptoms prior to age 14Psychotic symptoms among genetically low risk (GLR)Psychotic symptoms among genetically high risk (GHR)
None/few symptoms (NFS)Moderate symptoms (MS)Transitioned/psychosis (TP)None/few symptoms (NFS)Moderate symptoms (MS)Transitioned/psychosis (TP)
Male (n = 60)Female (n = 67)Male (n = 29)Female (n = 31)Male (n = 5)Female (n = 16)Male (n = 21)Female (n = 31)Male (n = 18)Female (n = 24)Male (n = 5)Female (n = 13)
Mean (SD)Mean (SD)Mean (SD)Mean (SD)Mean (SD)Mean (SD)Mean (SD)Mean (SD)Mean (SD)Mean (SD)Mean (SD)Mean (SD)
Depression0.45 (0.87)0.13 (0.38)0.17 (0.46)0.23 (0.62)1.40 (1.34)0.69 (1.19)0.48 (0.98)0.13 (0.43)0.72 (0.83)0.21 (0.51)0.40 (0.55)0.46 (0.77)
Anxiety0.17 (0.58)0.43 (1.06)0.41 (0.68)0.97 (1.43)1.00 (1.23)1.31 (0.95)0.62 (1.32)0.71 (1.10)1.28 (1.74)1.50 (1.45)0.20 (0.44)1.62 (1.56)
ADHD1.07 (2.09)0.61 (1.37)1.55 (1.62)1.10 (1.64)7.40 (5.32)2.62 (2.89)1.76 (2.74)1.19 (2.23)2.00 (1.94)2.00 (2.52)0.00 (0.00)3.00 (3.67)
ODD/CD0.77 (1.17)0.31 (0.63)0.69 (1.11)1.03 (1.25)4.40 (3.91)2.00 (1.97)1.14 (1.85)0.74 (1.36)1.22 (1.26)0.79 (1.02)1.60 (3.05)0.77 (1.01)
Marijuana Use0.43 (0.50)0.18 (0.38)0.10 (0.31)0.03 (0.18)0.20 (0.45)0.25 (0.45)0.34 (0.48)0.23 (0.42)0.33 (0.48)0.29 (0.46)1.00 (0.00)0.15 (0.38)
Mania0.12 (0.45)0.21 (0.64)0.31 (0.60)0.58 (1.15)0.20 (0.44)0.81 (1.17)0.43 (0.75)0.23 (0.49)0.11 (0.32)0.62 (1.09)0.40 (0.89)1.00 (1.68)
Delusions0.25 (0.54)0.37 (0.76)0.66 (0.86)1.29 (0.97)1.00 (1.00)2.38 (1.36)0.19 (0.41)0.35 (0.49)1.22 (1.26)1.54 (1.35)1.00 (1.23)2.00 (1.00)
Hallucinations0.43 (0.65)0.40 (0.84)1.69 (1.11)1.68 (1.08)2.60 (1.95)3.25 (1.81)0.29 (0.64)0.03 (0.18)1.61 (1.72)1.17 (1.34)2.80 (0.45)1.92 (1.38)
Age17.55 (1.18)17.12 (1.09)16.90 (1.23)16.87 (1.20)16.80 (1.64)16.88 (1.03)17.90 (1.18)16.87 (1.26)17.28 (1.13)17.33 (1.27)18.80 (2.59)17.23 (1.36)
Table 2.  Results of univariate analyses
Symptoms prior to age 14Gender, F(d.f. = 1)Genetic risk, F(d.f. = 1)Clinical status, F(d.f. = 2)Gender by genetic risk, F(d.f. = 1)Gender by clinical status, F(d.f. = 2)Genetic risk by clinical status, F(d.f. = 2)
  • *

    P < 0.05;

  • **

    P < 0.01.

  • a

    Males > females.

  • b

    Transitioned/psychosis > moderate symptoms and none/few symptoms.

  • c

    Genetic low-risk transitioned/psychosis > genetic high-risk transitioned/psychosis.

  • d

    Females > males.

  • e

    Genetic low risk > genetic high risk.

  • f

    Genetic low-risk males > genetic low-risk females.

  • g

    Genetic high-risk females > genetic high-risk males.

  • h

    Transitioned/psychotic males > transitioned/psychotic females.

  • i

    Genetic high risk > genetic low risk.

  • j

    Genetic high-risk transitioned/psychosis > genetic low-risk transitioned/psychosis.

  • k

    Transitioned/psychotic females > transitioned/psychotic males.

  • l

    Moderate symptoms > none/few symptoms.

  • m

    Genetic low-risk females > genetic low-risk males.

  • n

    Genetic high-risk males > genetic high-risk females.

Marijuana use12.39**a10.51**i3.90*b4.102.254.64*j

There were also significant effects for the interactions between gender by genetic status, F(8, 300) = 6.71, P < 0.01, gender by clinical status, F(16, 600) = 6.71, P < 0.01, and clinical status by genetic risk, F(16, 600) = 6.71, P < 0.01. Univariate analyses results, reported in Table 2, indicated that males in the GLR group recalled higher rates of ADHD symptoms than females in the GLR group. By contrast, females in the GHR group recalled higher rates of ADHD symptoms than males in the GHR group. Females in the GLR group recalled significantly higher rates of hallucinations than males in the GLR group; however, males in the GHR group recalled significantly higher rates of hallucinations than females in the GHR group. Males in the TP-group recalled more ODD/CD symptoms than females in that group. Females in the TP group recalled more mania and delusions than males in the TP group. Participants in the GLR group who were diagnosed with a psychotic disorder (GLR-TP) recalled more symptoms in the areas of depression, ADHD and ODD/CD compared with participants in the GHR group diagnosed with a psychotic disorder (GHR-TP). GHR-TP group participants reported higher rates of early marijuana use than GLR-TP group participants.


The current study focused on a culturally and genetically distinct sample of youth at genetically high and low risk for psychosis. Despite the limitation of using retrospective reports to identify early symptoms of psychosis, results contribute to the research literature in several respects. By including adolescents at different levels of genetic risk for psychosis, and including healthy, symptomatic and clinically disordered groups drawn from the same population of Palauan youth, we identified a complex set of early symptoms associated with prodromal psychosis and the occurrence of adolescent onset psychosis.

Although some previous researchers have suggested that there are rarely specific childhood precursor symptoms of a psychotic disorder that emerges in late adolescence or early adulthood,26 we found that young people in the MS and TP groups were more likely than their asymptomatic peers to recall psychotic symptoms in childhood. This is in line with previous research that has documented developmental continuity between symptoms and social and other impairments in childhood and the development of schizophrenia in adolescence.41,42 In particular, the results of our study reflect those of Poulton and colleagues who reported that self-reported psychotic symptoms at age 11 years predicted a very high risk of a schizophreniform diagnosis at age 26 years.43 Similarly, we found that the best predictor of adolescent onset psychosis among Palauans was the report of less severe psychotic symptoms experienced prior to 14 years of age. It is important to note that adolescents in the GLR-TP group also recalled symptoms across a broad spectrum of psychopathology, including depression, anxiety, CD/ODD and ADHD. This is in line with research suggesting that the earliest prodromal signs of an emerging psychotic disorder are non-psychotic and non-specific, with more specific, sub-clinical psychotic symptoms developing at a later stage (see Yung and McGorry44 for a review). Generally, these findings suggest that screening for psychiatric symptoms, particularly psychotic symptoms, could lead to the early identification of youth at high risk for psychotic disorders.

Not surprisingly, reported early symptoms of psychopathology varied between males and females. The increased reported experience of CD/ODD symptoms and marijuana use in males and increased rates of anxiety symptoms in females is similar to patterns previously observed by cross-cultural researchers.45 The higher rate of early depression scores among males is dissimilar to most epidemiological research that suggests that depression occurs more frequently among female children and adolescents. However, Rubenstein46 has previously reported that males between 15 and 24 years of age are the highest risk group for suicide in Micronesia (including Palau), and therefore, they may experience heightened rates of depression compared with females in the same age group. At the current time, it is more difficult to contextualize the finding that girls reported more early experiences of delusional and manic symptoms than boys. More psychiatric epidemiological research among children and adolescent is needed, particularly in developing nations, such as Palau.

More noteworthy were the findings that early mania and delusional symptoms were associated with the occurrence of a psychotic disorder among female adolescents, and that early marijuana use was associated with the occurrence of a psychotic disorder among male adolescents. These findings suggest that early indicators of adolescent onset psychosis may vary across gender, at least in Palau. This paper is the first to report a difference in the pre-psychotic experiences between males and females prior to the onset of psychosis. Other studies have either found no differences between male and female experiences47 or have not reported comparisons between the genders.15 A recent follow-up of the Edinburgh High Risk Study reported that men in that study who developed a psychotic disorder were more likely to report psychotic or possibly psychotic symptoms on entry to the study whereas women were less likely to report psychotic symptoms at baseline.21 It should be noted that participants in the Edinburgh study were significantly older than those in the current study and were not reporting childhood experiences of psychotic symptoms. Therefore, the studies are not directly comparable. Further research is required to determine if the gender differences in childhood experience reported in the current study are unique to Palau.

Gender differences in ADHD and hallucinations were found across the two genetic risk status groups. The finding that males in the GLR group were more likely to exhibit ADHD symptoms is consistent with much of the literature on gender and ADHD48. However, the finding that females in the GHR were more likely than males to report ADHD symptoms was unexpected. Other researchers have identified a link between attention problems and adolescent onset psychosis, but have not found this to be specific to males or females.49,50 The finding that females in the GLR group reported more hallucinations than males, but that males in the GHR group reported more hallucinations than males is difficult to interpret, as previous researchers have not noted such differences.

Perhaps most noteworthy were findings indicating that genetic risk status interacts with clinical status. GLR adolescents in the psychotic group (TP) were more likely to report early symptoms in the areas of depression, conduct problems and ADHD. This suggests that the occurrence of psychosis among Palauan youth, in the absence of a clear familial history, is associated with a broad range of early symptoms. Again, this reflects the broader literature describing the psychotic prodrome.44 Adolescents in the GHR group were more likely than their GLR counterparts to report early marijuana use, consistent with previous research on precursors of psychosis.26,51 These findings suggest that a substance abuse prevention program for genetically at-risk adolescents may help diminish the impact of psychosis, or at least delay its onset.

As indicated earlier, an important limitation of this study is that early symptom scores are based solely on self-reported recall. Recalling whether (and when) symptoms occurred at an earlier phase of life is a formidable challenge for many adolescents, let alone those who are experiencing psychotic symptoms. Although the reliance on memory was necessary to address our primary research question, we recognize that prospective studies of symptomatology use a superior methodology, and that such research is well worth pursuing. A related limitation is that many participants were quite young when the study was completed. Over time, it is likely that some of the participants will become more symptomatic, perhaps acutely psychotic, whereas others would become less symptomatic. Although we were able to identify a number of early indicators of psychotic disorder, it remains to be seen whether other early indicators will emerge as development unfolds.

The design of the current study and that of Woods et al.15 are very similar. Both compared a range of demographic, symptom and functional measures between young people thought to be at heightened risk of developing a psychotic disorder either due to clinical or genetic risk status and young people who did not meet heightened risk criteria. In both studies, significant factors that differentiated between the various groups were identified. The primary difference between the studies is that whereas Woods and colleagues assessed the experience of psychiatric symptoms and functioning at entry to their study, we asked participants to recall their experiences from earlier in childhood. Both studies suggest that individuals who are at risk of developing a psychotic disorder later in adolescence may experience difficulties prior to the emergence of psychotic symptoms.

The prevalence of psychotic symptoms among a community-based sample of Palauan youth in both the GHR and the GLR groups raises the possibility that this isolated sample is both culturally and genetically unique. In addition to having a high prevalence rate of untreated psychosis and sub-clinical psychosis among adolescents in genetically affected families, it appears that the prevalence rates of prodromal psychosis and functional psychosis among adolescents from unaffected families was also quite high.52 This finding was somewhat unexpected, as it was assumed that psychosis in Palau would be primarily clustered within genetically high-risk families. The large proportion of Palauan adolescents who report psychosis allowed us to identify a number of early symptoms, but the somewhat unusual patterning of these symptoms suggests that the nature of psychosis observed among Palauan youth may be different from psychosis observed in Western nations.


This study was supported with funds from NIMH grant number MH5418 and Fogarty International Center Grant number TW007803.