We investigated the psychometric properties of the Portuguese version of the CAPE-42 and the association between childhood adversity, cannabis use and the occurrence of PEs in Brazilian individuals from the general population. The Brazilian version of the CAPE presented acceptable reliability and validity after removing nine items, maintaining the factorial distribution proposed in the original instrument3 for all the three dimensions: positive, negative, and depressive. PEs were detected in individuals from the Brazilian community and were significantly associated with heavy cannabis use and childhood adversity.

Although the total sample size (n=640) met the requirements for CFA,26 good adjustment indexes were not initially obtained the original CAPE model. An EFA was then performed26 and nine items did not obtain a good factorial load (> 0.50), seven (items 5, 6, 11, 13, 17, and 20) from the positive dimension and two (items 4 and 8) from the negative dimension.

One possible explanation for the low scores on these items is that terms used for the Capgras syndrome and Voodoo items, for example, might have been unfamiliar to the participants, which could have led to confusion about their meaning. Thus, a high frequency of never responses occurred for these items. According to the Index of Basic Education Development,28 Brazilian education levels are still below expectations for emerging countries. An alternative hypothesis is that these items describe ideas and behaviors that are culturally misjudged by the sample, particularly those related to religious or mystical content, which may also have led to inconsistencies in the responses. Despite Brazil’s cultural diversity and religious syncretism, Christianity is the largest religion, with a high number of Catholics (64.6%) and Protestants (22.2%) citizens (http://www.ibge.gov.br).

Item 6 (False appearance) may have suggested a double meaning and thus failed to explain the construct it was intended to represent. In the Spanish version of the CAPE, the same item did not obtain a good factorial load and was reallocated to the depressive dimension.14 The same misinterpretation could have occurred in items 4 (Not talkative) and 8 (No emotion). In the German version of the CAPE, for example, items 4 and 8 were made subfactors in the negative dimension (emotional withdrawal and affective blunting, respectively), which allowed better factor loads and good adjustment indices.15

After the nine items were removed, acceptable adjustment indices were found in CFA of the Brazilian CAPE-33, which were distributed in three dimensions: 13 positive, 12 negative, and eight depressive. The absolute (CFI = 0.895) and relative (GFI = 0.822) indices were very close to good quality (> 0.9). In parsimony analysis, the instrument obtained a PGFI = 0.761 and, in the population discrepancy, the adjustment was acceptable (RMSEA = 0.055), being close to the reference value (≤ 0.05).26

The internal consistency analysis of the CAPE-33 resulted in good Cronbach's α values in the three dimensions. The negative dimension obtained the highest value (α = 0.89), close to that found in a Spanish study,14 in which psychotic patients were assessed (α = 0.88), as well as that found in an Italian study,13 which included a sample of non-psychotic students (α = 0.86). The German version of the CAPE found the same internal consistency value in the positive dimension as the Brazilian CAPE-33 (α = 0.88) and a very close value in the depressive dimension (α = 0.85 for the German version and α = 0.86 for the Brazilian version).15 This agreement could be due to the characteristics of the samples, since both studies included people from the general population and psychosis patients.

The distribution of the FEP sample by sex was similar to that described in epidemiological studies,18 with a higher percentage of PEs observed in men. The severity of the disorder was associated with lower educational levels and poor social functioning.29

As expected, the patient group reported PEs at a greater frequency than the unaffected siblings group and the population-based group. Although we cannot rule out the possibility of under-reporting by the siblings, we do not believe that this occurred, first of all because the included siblings were possibly healthier, and perhaps the women were caregivers. In addition, in all dimensions, there were siblings and population-based participants with high CAPE-33 scores, similar to those observed in the patient group. These participants, who varied from expected group values, are called outliers in statistical terms, meaning atypical values that are far from standard values for a given sample. The presence of atypical values in CAPE-33 scores among individuals undiagnosed with a psychotic disorder corroborates the hypothesis of PE continuity in the general population, reinforcing the proposition that psychosis is not limited to categories described in diagnostic manuals, such as the DSM-5.

Reports of PEs in the negative and depressive dimensions were greater among women than men. This could be associated with a higher prevalence of depressive symptoms in women and, thus, a predisposition to report negative and depressive experiences.30 In an Italian study,13 for example, although the authors indicated that no substantial differences were found between the sexes, mean CAPE scores were higher in all dimensions for women than men.

We found no association between PE and age range, marital status, or education level. This result seems to go against previous data suggesting that PEs would be more common in younger, less educated, and single people.5,31 This could indicate that PEs are characterized differently in emerging countries than in countries with higher incomes, where such studies have predominated.32

In our Brazilian community-based sample, the experience of any childhood adversity (physical, emotional, or sexual abuse, or physical or emotional neglect) was positively associated with PEs in all CAPE-33 dimensions. This agrees with what has been reported in previous studies on non-clinical populations.33,34 Another important finding was the association of two or more types of childhood adversity with greater reports of PEs in all CAPE-33 dimensions. This result could be related to the association between higher frequency, persistence, and intensity of exposure to early stress with worse mental health outcomes, which corroborates previous findings.35

The association between PE and cannabis use was significant only in the positive dimension, which was similar to the findings of studies conducted with non-clinical populations in other countries.36 This reinforces the fact that the higher the substance use, the greater the rate of PEs in the studied population.37

This study has some limitations. First, conducting a pilot study to test the semantic quality of the questions could have helped reformulate items later considered of low comprehension. Second, we were unable to test the effects of specific subtypes of abuse and neglect and their association with PEs due to the relatively scarcity of participants in the population-based group who reported each adversity subtype. In addition, the CAPE does not collect temporal data on PE onset, and thus we were unable to refine our statistical analysis by including volunteers whose PE onset was prior to the occurrence of childhood adversity or initiation of cannabis use. Regarding the inclusion of our volunteers, no randomization method was possible in sibling recruitment, which could have led to a bias towards recruiting healthier siblings. It is important to point out that the included siblings might have been the patients’ caregivers, and for this reason agreed to take part in the study.

Despite these limitations, our study has a number of strengths. Firstly, the non-psychotic group, consisting of population-based participants residing in the epidemiologic catchment area, ensured greater representativeness of the general population.38 This approach also strengthened the search for further psychometric evidence of the CAPE instrument, as well as characterization of manifestations in the psychosis spectrum, in the general population.4 Secondly, using a standardized instrument to confirm the psychiatric diagnosis (SCID for DSM-IV) ensured the exclusion of participants with psychotic disorders from the general population. Finally, this is the first study to validate a version of the CAPE in an emerging country, which has allowed investigation of PEs in a Brazilian sample. Our study will certainly contribute to future epidemiological studies aiming to understand the psychosis spectrum in developed and emerging nations.

In conclusion, the Brazilian version of the CAPE presented acceptable psychometric properties after excluding nine items. In a population-based sample, PEs were reported by people not diagnosed with any psychotic disorder. Likewise, PEs were associated with common adversity, such as childhood adversities, and cannabis use in individuals from the community, which is similar to what has been described in other countries and cultures and reinforces the validity of the CAPE. Given that PEs are more prevalent than psychotic disorders, this phenomenon could represent an intermediate step towards a better understanding of psychoses. Further studies that seek associations between PEs and other social and biological risk factors could contribute to a deeper understanding of the etiology and physiopathogenesis of psychoses, as well as the risk and protective factors for psychoses in emerging countries. Finally, it would be interesting to compare the CAPE’s psychometric performance with other instruments validated in Brazil, e.g., the Composite International Diagnostic Interview,39 the Positive and Negative Syndrome Scale, the Schizotypal Personality Questionnaire,40 and the Magical Ideation scale,41 which could be used to estimate the occurrence of PE-related constructs.