Racial and Ethnic Differences in the Prevalence of Psychotic Symptoms in the General Population

Psychotic symptoms present in milder forms in the general population, with worldwide estimates of .8%−31.4% and a median prevalence of 5.3% (1,2). Although psychotic symptoms may be common, questions remain regarding whether these symptoms are benign, a sign of distress, or indicative of a mental disorder. Several investigators have contended that psychotic symptoms are on a continuum, supported by evidence that the symptoms are frequent but the disorders are not (3–7) and by statistical models that suggest a dimensional rather than a categorical structure (8). Both psychotic symptoms and psychotic disorders have been linked to similar variables, such as female gender, younger age, being unmarried, urban residence, low income, drug use, trauma, social dysfunction, depression, and neurotic symptoms (9–13).

Because psychotic symptoms may serve as precursors to psychotic disorders, examining racial-ethnic differences in psychotic symptoms could be instructive with respect to causal mechanisms of schizophrenia and the impact of racial-ethnic disparities. In the United Kingdom, epidemiological studies have consistently found higher rates of schizophrenia among blacks of Caribbean and African descent (14). In addition, in the United States, the Epidemiologic Catchment Area study found lifetime prevalence of schizophrenia to be significantly higher among blacks (2.1%) than among whites (1.4%) or Hispanics (.8%) (15). The first National Comorbidity Survey (NCS-1) found higher odds ratios (ORs) for nonaffective psychotic disorders among blacks (OR=1.9), Hispanics (OR=1.5), and other groups (OR=1.9) compared with whites, although these differences were not statistically significant (7), whereas the NCS-2 found lower odds ratios for nonaffective disorders among blacks (OR=.8) and Hispanics (OR=.7) compared with whites, although these differences were also not significant (16). In a prospective study of birth cohorts in the United States, Bresnahan and coauthors (17) found that after the analysis controlled for family socioeconomic status, African Americans were still twice as likely as whites to develop a psychotic disorder. Variations in the relative rates of psychotic disorders most likely reflect limited statistical power and methodological differences between studies.

In contrast to studies on schizophrenia, there is a paucity of epidemiological data regarding psychotic symptoms and race-ethnicity. Several studies in the United Kingdom found higher rates of psychotic symptoms among black Caribbean and African populations compared with whites (9,12). A study conducted in a large, urban primary care practice in the United States found a high correlation of psychotic symptoms with Hispanic ethnicity (10). It is unclear to what extent these racial-ethnic differences reflect other associated psychosocial factors or cultural expressions (“idioms”) of distress (12,18,19).

Because of its diverse racial and ethnic communities, the United States provides an excellent laboratory to examine the association between culture and psychotic symptoms. However, previous investigations of psychotic symptoms in the general community have been limited by sampling problems (for example, representativeness and design effects); failure to examine the effects of confounding variables; and inattention to the relationship between symptoms, psychological distress, and daily functioning. In this study we drew on data derived from the Collaborative Psychiatric Epidemiology Surveys (CPES) (20) to address these limitations. The CPES consists of three nationally representative surveys—two of which focused specifically on minority populations—that have been adjusted for design effects. Thus, using the CPES database, this study had the following objectives: to determine the prevalence of psychotic symptoms among various racial-ethnic groups in a nationally representative sample of adults in the United States and to explore the relationship of psychotic symptoms, race-ethnicity, psychological distress, and dysfunction with respect to the issues of psychiatric disorders, idioms of distress, and everyday phenomena.

Methods

Survey data

Data were derived from the CPES (20), which combines three nationally representative surveys: the National Comorbidity Survey Replication (NCS-R), the National Survey of American Life (NSAL), and the National Latino and Asian American Study (NLAAS). These three surveys were designed a priori to be merged, and they used the same core diagnostic questions. The NCS-R is a replication of the NCS-1. It was conducted from 2001 to 2003 and yielded core diagnostic assessments of all 9,282 survey respondents in part 1, as well as a subset of 5,692 respondents in part 2, which focused on risk factors, consequences, correlates, and additional disorders. The NSAL explored racial and ethnic differences in mental illness among whites, African Americans, and Afro-Caribbeans. It was conducted from 2001 to 2003 and yielded 6,082 respondents in the CPES. The NLAAS focused on Latino and Asian Americans, was conducted in 2002–2003, and had 4,864 respondents.

All surveys were conducted among adults age 18 and older who were living in the contiguous United States; all surveys used laptop computer–assisted personal interviews in the homes of respondents. Initially, each study was divided into subgroups by race-ethnicity and geographic region. The combined-analysis weights were created by using the adjusted study-specific weights that were scaled by relative size of the study subgroup to the combined subgroup across all three studies. Thus the combination of these studies provided sufficient power to investigate cultural and ethnic influences on mental disorders as though the studies were a single nationally representative study.

The core CPES questionnaire was used in all three studies and was derived from the World Health Organization's (WHO) expanded version of the Composite International Diagnostic Interview (CIDI) developed for the World Mental Health (WMH) Survey Initiative, the WMH-CIDI (21). Sociodemographic and clinical variables included in the survey and used in the analysis were age, gender, race-ethnicity, country of birth, marital status, education, income, region of the United States, and the presence of a substance use disorder and posttraumatic stress disorder (PTSD).

Psychotic symptoms

Delusions were defined as respondents’ endorsement of the experience of any of the following over their lifetime: mind control, mind taken over by strange forces, communication attempts by strange forces, or “unjust plot to harm you and being followed, but family and friends do not believe you.” Hallucinations were defined as seeing visions or hearing voices that others could not see or hear at any point in one’s lifetime. To account for possible recall biases, we also examined 12-month prevalence of psychotic symptoms, which was available as a single item based on whether respondents had experienced any of the above symptoms in the previous 12 months.

Psychological factors and disability

Psychological distress was defined as experiencing two or more of the following in one’s lifetime: being sad, empty, or depressed for a period of several days; being discouraged about life for a period of several days; losing interest in enjoyable things for a period of several days; having a sense that life is meaningless for a period of several days; feeling irritable, grumpy or in a bad mood for a period of several days; worrying more than others about the same problems; feeling more nervous or anxious than others about same problems; or feeling anxious or worried for most days for one month or more. Dysfunction was defined as having three or more days in the previous month during which normal daily activities were limited because of problems with physical or mental health. Three days was used as the cutoff because it exceeded the median value of two days for the general population. Diagnoses of lifetime alcohol and drug abuse and dependence and lifetime PTSD were based on DSM-IV criteria (22).

Statistical methods

To address the first objective—to determine the prevalence of psychotic symptoms in various racial-ethnic groups—we used a complex sample design and weighted estimates developed for the combined sample from the three studies (N=16,423). Chi square analysis was used to compare the groups. To address the second objective—to elucidate the relationship between psychotic symptoms and race-ethnicity, psychological distress, and dysfunction—bivariate and multivariate analyses were performed with chi square and multivariate logistic regression, respectively. The dependent variables were lifetime and 12-month prevalence of psychotic symptoms. The independent variables were the race-ethnicity, distress, and disability variables, and the other demographic and clinical variables served as covariates. Missing responses were not analyzed. A p value of <.05 was considered statistically significant. All analyses were performed with SPSS, version 18. Institutional review board approval was not necessary because the analysis was of deidentified national data.

Results

Demographic characteristics

As shown in Table 1, more than two-thirds of the sample was white, and the next largest groups were Latino and black Americans. Among African Americans, an overwhelming majority (94%) were born in the United States, with the remainder being primarily of Afro-Caribbean descent. Nearly three of five participants were married, and almost one in five had not completed high school. One-quarter of the sample had an annual household income of less than $22,500. The proportion of females (53%) was slightly higher than the proportion of males (47%), and the largest proportion of the sample (34%) was living in the South.

Prevalence of psychotic symptoms

The adjusted lifetime prevalence rates of delusions and hallucinations in the study sample were 2.2% and 10.7%, respectively, and 11.6% of the respondents (N=1,593) had experienced a psychotic symptom in their lifetime. The unadjusted 12-month prevalence of psychotic symptoms was 2.3% (N=455), which was 29% of those who reported having experienced a psychotic symptom in their lifetime. After adjustment for design effects, the 12-month prevalence declined to 1.4%.

Bivariate analysis

There were significant racial-ethnic differences among persons who reported experiencing psychotic symptoms (Table 1). Latinos and blacks had a higher lifetime prevalence of psychotic symptoms (13.6% and 15.3%, respectively) compared with whites (9.7%) and Asians (9.6%). No racial-ethnic differences were found between persons who reported experiencing only delusions. No significant association was found between black ethnicity (African American or Afro-Caribbean) and psychotic symptoms (data not shown). With respect to other social and clinical variables, the prevalence of psychotic symptoms was higher among women, persons with less than a high school education, those with annual household incomes below $22,500, those residing in the West, and those with a lifetime DSM-IV diagnosis of substance abuse or dependence and a lifetime history of PTSD compared with other groups in the respective demographic or clinical categories. No association was found between country of origin and psychotic symptoms (data not shown).

Nearly half of all respondents (45%) reported having experienced psychological distress in their lifetime (Table 2). Psychological distress was associated with a twofold higher prevalence rate of hallucinations and lifetime psychotic symptoms and a sixfold higher prevalence rate of delusions. Similarly, among persons who reported three or more days of limited daily activities in the past month, the prevalence of lifetime hallucinations or any psychotic symptoms was twice as high and the prevalence of lifetime delusions was about 2.5 times as high as among those who reported none or fewer days of limited activity.

Among persons who reported having experienced psychotic symptoms in their lifetimes, 21% had no history of distress or dysfunction, 48% had distress but no dysfunction, 6% had dysfunction and no distress, and 25% had dysfunction and distress. Among persons with no history of psychotic symptoms, 45% had no history of distress or dysfunction, 39% had distress but no dysfunction, 4% had dysfunction and no distress, and 12% had dysfunction and distress. The differences between respondents who did and did not experience psychotic symptoms were significant for both distress (χ²=241.51, df=1, p<.001) and dysfunction (χ²=95.65, df=1, p<.001). The lifetime prevalence rates of psychotic symptoms with distress and without distress, respectively, for each of the racial-ethnic groups were as follows: Asian, 5.4% and 6.4%; Latino, 19.9% and 8.2%; black, 21.1% and 9.9%; and white, 13.1% and 5.1%.

The 12-month prevalence data indicated significant associations between psychotic symptoms and various demographic variables–race-ethnicity, age, marital status, education, and income (Table 1)—and between psychotic symptoms and various clinical variables, such as distress, lifetime diagnosis of a substance use disorder, and lifetime diagnosis of PTSD (Table 2).

Logistic regression analysis

In logistic regression analysis (Table 3), lifetime prevalence of psychotic symptoms was significantly associated with the three independent variables: race-ethnicity (Latino), psychological distress, and dysfunction; it was also associated with two covariates: a lifetime diagnosis of a substance use disorder or of PTSD. Notably, experiencing high levels of distress was associated with higher odds (2.1 times higher) of lifetime psychotic symptoms, and having three or more recent days of dysfunction was associated with higher odds (1.6 times higher) of lifetime psychotic symptoms. Several variables that had been significant in the bivariate analysis—gender, age, education, income, and black race—were no longer significant.

Logistic regression analysis with 12-month prevalence data (Table 3) indicated that reporting psychotic symptoms in the past 12 months was associated with three independent predictors—being Latino, being black, and experiencing higher distress—and two covariates—being unmarried and having a lifetime diagnosis of PTSD.

Discussion

To our knowledge, this is the largest epidemiological study of the prevalence of psychotic symptoms in various U.S. racial-ethnic groups. Our finding that 11.6% of the sample reported having experienced psychotic symptoms in their lifetime is consistent with results of several other large-scale studies. The NCS-R study found that 9.1% of respondents endorsed items for lifetime symptoms of nonaffective psychosis, with 6.3% of respondents reporting visual hallucinations and 4.0% reporting auditory hallucinations (16). A Dutch study using the CIDI questionnaire found a lifetime prevalence of psychotic or psychosis-like symptoms of 17.5% (11).

Our 12-month prevalence data are more perplexing because the unadjusted and adjusted prevalence rates, 2.3% and 1.4%, respectively, are below most estimates reported in other studies. A meta-analysis by van Os and colleagues (2) found a median prevalence rate of 5.3%, although a few studies had 12-month prevalence rates as low as 1%. A WHO study estimated a 12-month symptom prevalence range of .7%−45.8% with rates of 3.3%−16.3% in countries at high and upper-middle economic levels (1). However, three of the countries of the upper-middle economic level had prevalence rates below 2% when slightly more strict criteria were used. It is possible that the relatively low 12-month prevalence in our study reflected the derivation of the psychotic symptom variable from a single item, whereas other studies commonly used responses to several items. The latter strategy was used to create our lifetime measure.

With respect to the relationship between race-ethnicity and psychotic symptoms, after our analysis controlled for various clinical and sociodemographic factors, only Latinos had significantly higher odds of reporting psychotic symptoms. Higher prevalence among blacks disappeared after control for other variables, although the OR for psychotic symptoms among blacks (2.0) was the highest among all racial-ethnic groups. Moreover, the 12-month prevalence data indicated that blacks as well as Latinos had significantly higher rates of psychotic symptoms than whites. Our lifetime results differed from those of a British study that found a significantly higher rate of psychotic symptoms among black Caribbeans than among whites after adjustment for potentially confounding demographic variables (11). However, our lifetime findings were partially consistent with a second British study that found that after control for social disadvantage, differences were reduced between Afro-Caribbeans and whites but not between persons of African descent and whites (12).

Our findings for lifetime prevalence resemble those from an urban, general medical practice in the United States in which Hispanics had the highest prevalence of psychotic symptoms (10). Several investigators have underscored the importance of recognizing the cultural and environmental context of psychotic symptoms (10,12). In writing about “culturally patterned idioms of distress,” Olfson and coauthors (10) cautioned that it may be difficult to distinguish true psychotic symptoms from culturally or religiously sanctioned understandings of dissociative or supernatural processes. However, we found that, with the exception of Asians, prevalence rates of psychotic symptoms were roughly twice as high for persons who had a history of distress regardless of racial-ethnic group. This suggests that distress and psychotic symptoms have a more general association and may not be affected substantially by culture. Nevertheless, there may be some groups, such as Asians, in which psychotic symptoms are not typically associated with distress. Thus, contrary to earlier views concerning psychosis as a cultural idiom of distress (23), culture may be more relevant as an exception to the general tendency linking psychosis and distress.

Our findings adumbrate several important issues with respect to the transition from psychotic symptoms to clinical diagnosis. The propensity for blacks to be more apt than whites to receive a diagnosis of schizophrenia is consistent with our finding that a larger proportion of blacks than whites reported lifetime psychotic symptoms. However, after the analysis controlled for various confounding variables, the racial differences in psychotic symptoms were attenuated, although the differences in 12-month prevalence rates remained significant. This suggests that an appreciable amount of the variance accounting for the diagnosis of schizophrenia among blacks may be secondary to other factors. Indeed, Bresnahan and coauthors (17), controlling for only family socioeconomic status, were able to reduce the OR among blacks for receiving a diagnosis of schizophrenia by 50%. On the other hand, although a larger proportion of Latinos than whites in our study reported psychotic symptoms, their likelihood of being diagnosed as having schizophrenia did not seem to be greater than whites. In a large national study, Lewis-Fernández and coauthors (24) found that only 7% of Latinos who endorsed psychotic symptoms met DSM-IV criteria for a psychotic disorder. Thus there is considerable complexity in the transition from self-reports of psychotic symptoms to a diagnosis of a psychotic disorder, which must include a consideration not only of the impact of clinical and sociodemographic factors but also of the role of labeling issues, racial-ethnic disparities, and biological factors; biological factors were not examined in this study.

Finally, our findings raise questions as to whether psychotic symptoms should be part of screening by primary care physicians and whether such symptoms require intervention. It has been found that individuals who screen positive for psychotic symptoms have increased rates of psychiatric hospitalization and mental disorders, and the risk of developing a psychotic disorder rises with more self-reported psychotic symptoms (25). Loewy and coauthors (26) stated that screening for psychotic disorders needed to take into account stigma and resource allocation as well as increased specificity, and that such screening may be more appropriate among persons who report distressing or frequent psychotic symptoms. Although we could not determine a direct temporal relationship, we found that psychotic symptoms were reported more frequently by persons who also reported histories of distress or dysfunction and that only one-fifth of persons with a history of psychotic symptoms had no history of distress or dysfunction. Our 12-month data also confirmed the strong association between distress and psychotic symptoms, although the association with dysfunction did not obtain. Thus a history of psychotic symptoms may serve as an indicator of enhanced risk for other psychological difficulties. This viewpoint is consistent with those of Lewis-Fernández and colleagues (24), who concluded that endorsement of psychotic symptoms by Latinos “may constitute a clinically significant marker of general psychiatric vulnerability rather than a sign of psychotic disorder.”

This study had several strengths, including a large, multiracial, representative population-based sample of persons in the United States that is adjusted for various design effects and a multivariable analysis that allowed for the assessment of the unique associations between various predictor variables and psychotic symptoms. The study also had several limitations. First, symptoms of psychosis were assessed by lay interviewers and were self-reported, which makes these variables more susceptible to false positives, especially with respect to clinical significance. The NCS-R subanalysis of persons with nonaffective psychosis disorder who had endorsed items used in our study found that 16.8% were probable cases of nonaffective psychosis, and the lifetime estimate of true nonaffective psychosis was 1.5% (16). Second, with the exception of dysfunction, which was assessed for the previous month, many variables were assessed over a respondent’s lifetime and were subject to recall bias; this also limited our ability to determine their association with lifetime psychotic symptoms. Thus the relationships between these variables and psychotic symptoms suggest associations but not a direct temporal correspondence. It was reassuring that the 12-month findings, which might be subject to less recall bias, were largely consistent with the lifetime data. However, the 12-month data must be interpreted cautiously because they were derived from a small sample. Third, the variables for distress and dysfunction are proxy variables and are not based on established scales. Finally, the data were cross-sectional and causal relationships could not be determined.

Conclusions

This study found that race-ethnicity was differentially associated with psychotic symptoms, with Latinos more likely than other groups to report lifetime psychotic symptoms after the analysis controlled for various confounding variables. Persons with a history of psychotic symptoms were prone to higher rates of distress, and this association was not specific to any racial-ethnic group. Thus psychotic symptoms do not typically represent cultural idioms of distress in the United States. Although psychotic symptoms are often transient (23), their presence at any point over the life course can serve as an indicator of a propensity to experience psychological distress.