Turner RJ, Gil AG. Psychiatric and Substance Use Disorders in South FloridaRacial/Ethnic and Gender Contrasts in a Young Adult Cohort. Arch Gen Psychiatry. 2002;59(1):43-50. doi:10.1001/archpsyc.59.1.43
Prevalence rates of psychiatric and substance use disorders among young adults in South Florida are presented. Unique aspects of the study include the large sample size, its ethnic diversity, and the fact that a substantial proportion of Hispanic participants were foreign born.
This study builds on a previous cohort study of students who entered middle school in 1990. A random subsample of this representative cohort (N= 1803) was interviewed between 1998 and 2000 when most were between 19 and21 years of age. Disorders were assessed through computer-assisted personal interviews utilizing the DSM-IV version of the Michigan Composite International Diagnostic Interview.
More than 60% of the sample met lifetime criteria for 1 or more study disorders, and 38% did so within the preceding year. Childhood conduct and major depressive and alcohol abuse disorders were the most prevalent. Although rates of affective and anxiety disorders in females were double that in males, this gender difference disappeared when attention-deficit/hyperactivity disorder, conduct disorders, and antisocial personality disorders were also considered(46.6% vs 45.7% for females vs males, respectively). Substantially lower rates were observed among African Americans for depressive disorders and substance abuse and dependence. Among Hispanics, rates tend to be lower among the foreign-born in comparison with their US-born counterparts, particularly for the substance disorders.
The documented presence of psychiatric and substance disorders in middle and high school populations emphasizes the importance of prevention efforts in school settings. Research on the origins of ethnic and nativity differences is called for.
CREDITABLE information on the community prevalence and demographic correlates of psychiatric and substance use disorders has been available in published form for little more than a decade. Based on data obtained from 5 separate and largely urban sites, the Epidemiologic Catchment Area Study (ECA) provided estimates of lifetime and current psychiatric and substance use disorders within and across African American, Hispanic, and non-Hispanic white subpopulations.1 Subsequently, the National Comorbidity Survey (NCS)2 employed a nationally representative sample of more than 8000 persons between the ages of 15 and 54 to provide estimates of the prevalence of psychiatric and substance use disorders and their social and ethnic distributions. A summary of the ECA and NCS findings has recently been presented by Tohen et al.3 Recent advances have also been made toward estimating the community burden of substance use and psychiatric disorders among children and adolescents 9 to 18 years of age.4- 7
These studies have been of immense theoretical and practical significance because they have (1) produced estimates of the true prevalence of specific psychiatric disorders in the community using well-defined diagnostic criteria;(2) provided fundamental information on the chronicity, course, and comorbidity of psychiatric disorders; (3) provided a basis for estimating the extent and nature of unmet service needs; and (4) identified subpopulations most in need of, or who might benefit most from prevention efforts.
This article is based on a new study within this tradition. It presents findings on the prevalence and demographic distributions of psychiatric and substance use disorders among a representative cohort of 1803 young adults. Most participants (93%) were between 19 and 21 years of age when interviewed between 1998 and 2000. The results constitute a unique contribution in several respects. First, we believe that these data are from the largest sample within this age range so far studied in the United States. Second, this is among the first large-scale community studies to estimate prevalence rates based on DSM-IV criteria. Most significantly, half of this sample is composed of the understudied and quite distinct Hispanic population of South Florida. Specifically, the sample was drawn such that approximately25% were of Cuban origin, 25% were other Caribbean basin Hispanic, 25% were African American, and 25% were non-Hispanic white. In addition, a substantial proportion of Hispanic participants were foreign born (44.5%), a factor that has been shown to be relevant to mental health and substance abuse risk.8- 10
Our approach in drawing this sample is in accord with a growing consensus in the field that race is more of a social than a biological categorization akin to ethnic status,11 and that there are important cultural variations within ethnic statuses. In an effort to minimize the effects of such variations on results, we have distinguished Cubans from other Hispanics and limited inclusion in the latter category to Hispanics from countries in the Caribbean basin. For the same reason, Haitians and other Caribbean black participants were excluded in forming the African American subsample. Because of our interest in the effects of immigration and immigration status, we have also excluded Puerto Ricans from the "other Hispanic" category.
This study builds on a previous 3-wave investigation based in the Miami-Dade public school system.8 All 48 of the county's public middle schools and all 25 public high schools as well as alternative schools had participated. Questionnaires were administered annually between1990 and 1993 beginning in grades 6 and 7 and finishing when participating students were in grades 8 and 9. Consent forms were sent to parents of the total population of 9763 male students scheduled to enter sixth and seventh grades, and of 669 female students from 6 schools selected to approximate the overall ethnic composition of county middle schools. Of these 10 432 prospective participants, completed questionnaires were obtained from 7386 at wave 1, from 6646 at wave 2, and from 5924 at wave 3. Detailed analyses provided assurance that wave 1 participants were highly representative of the population from which they were drawn and that this was also true for the wave 3 participants, despite a nearly 20% attrition across the 3 data points.8
Within the confines of ethnicity criteria, all female participants in the earlier investigation (n = 410), and a random sample of 1273 male participants, were ultimately selected for follow-up. Because a relatively small number of girls were included in the parent study, a supplementary sample was randomly drawn from the Miami-Dade county 1990 sixth- and seventh-grade class roster. Figure 1 summarizes the results of fieldwork efforts. Overall, 70.1% of those we searched for and attempted to recruit to the study were successfully interviewed. By far, the greatest loss occurred among the new sample of girls who had no involvement in the early-adolescent study. A success rate of 76.4% was achieved among those in the original sample, despite the fact that many had left home for college or for other reasons.
Those interviewed were compared with the total sample of individuals drawn from the original study population on 28 early-adolescent behaviors and family characteristics of possible relevance to mental health or substance abuse risk (analyses not shown). No statistically significant differences were observed. Comparisons were also made with respect to school dropout. Among those interviewed, 20.5% reported that they had dropped out of high school. This corresponds closely with rates reported by the school board on the same student cohort of 21.1% for boys and 15.2% for girls (data available at: http://www.dade.k12.fl.us/eema/Abstract%202000/Abstract_2000_Site/index.htm, accessed November 28, 2000). These comparisons and the 76.4% follow-up success rate allow the conclusion that our sample is representative of the population from which it was drawn. In contrast, the 58.2% success rate with the supplementary sample of new girls is disappointing, and analyses revealed a significant parental socioeconomic status bias associated with these losses. To correct for this bias, female participants have been differentially weighted in all analyses to achieve a distribution on socioeconomic status that approximates that observed for male participants. Because we sampled so as to achieve roughly equal numbers of white non-Hispanic, Cuban, other Hispanic, and African American participants (except where results are presented by ethnicity), the data have also been weighted to population values with respect to ethnicity and gender.
Data on the lifetime prevalence and 1-year prevalence of psychiatric and substance use disorders were obtained through computer-assisted personal interviews that allowed estimation of DSM-IV diagnoses. Our basic instrument was the Michigan Composite International Diagnostic Interview(CIDI), which was employed in the NCS.2 The CIDI is a fully structured interview based substantially on the Diagnostic Interview Schedule (DIS)12 and designed to be administered by nonclinicians trained in its use.13,14 Using the Michigan CIDI, as updated by NCS researchers to cover DSM-IV criteria, we assessed major depression, dysthymia, generalized anxiety disorder, social phobia, panic disorder, alcohol abuse and dependence, drug abuse and dependence, posttraumatic stress disorder (PTSD), and antisocial personality disorder. The latter 2 modules had been borrowed from the DIS12 for the NCS. Field trials of the original CIDI had documented good reliability and validity for all of the CIDI diagnoses considered here.15 Evidence for the validity of the Michigan CIDI diagnostic estimates, evaluated against structured clinical reinterviews,16 have been reported for most NCS disorders, including affective disorders,17 anxiety disorders,18,19 addictive disorders,20,21 and posttraumatic stress disorder.22
Along with the Michigan CIDI, our assessment instrument23 included a reliable module24 taken from the revised DIS25 to assess attention-deficit/hyperactivity disorder (ADHD) and items to allow assessment of childhood conduct disorder. The NCS strategy of a preliminary screening process was extended to also include the lifetime use of individual licit and illicit drugs. The goal of this extension was to reduce any falloff in reporting that might be occasioned by learning during the course of the interview that positive responses, and not negative responses to drug questions, tend to be followed by a large battery of additional questions. Finally, our procedure with the PTSD module differed from both the NCS and the ECA1 studies. Following an extensive battery of 41 questions on major and potentially traumatic experiences, we followed the standard procedure of administering PTSD questions in relation to the event nominated by respondents as being the worst. However, when, a diagnostic criterion was not met, participants were presented with a list of the major PTSD symptoms and asked whether they had ever experienced any of them in relation to any other event. If a participant responded "yes," the event was specified and the PTSD module repeated. This is an efficient procedure that effectively minimizes the risk of false negatives.
All interviewers held bachelors degrees and most of them had some graduate education. They were given a total of 7 days of training, 2 days on general interviewing techniques and procedures, and 5 days on the CIDI. Except for the initial cohort, this training was followed by the observation of 2 interviews conducted by experienced interviewers, and by being observed while conducting interviews of their own. The use of laptop computers assured appropriate skip patterns and greatly facilitated the reliable administration of the interview. Our standard practice was face-to-face interviewing in the respondent's home or in our research offices as the respondent chose. However, telephone interviews using previously mailed response booklets were employed for those who were away at university or who had moved elsewhere in the contiguous United States. Approximately 30% of the interviews were conducted by telephone. There is abundant evidence that in-person and telephone interviews yield comparable data.26- 28
Analyses were conducted using SPSS 10 (Statistical Products and Service Solutions 10; SPSS Inc, Chicago, Ill). As noted above, data were weighted to correct for an underrepresentation of girls with higher socioeconomic status and to reflect population distributions on gender and ethnicity. The CROSSTABS and DESCRIPTIVES programs of the SPSS software package were used to compute χ2 tests of the significance of prevalence differences across gender and ethnicity.
Examination of lifetime and past-year prevalence revealed that more than 60% of the sample met criteria for 1 or more study disorders at some time during their lives, and 38% did so during the 12 months preceding the interview (Table 1 and Table 2). Childhood conduct disorder shows the highest lifetime prevalence, followed by major depressive disorder and alcohol abuse. It is clear from these results that the consistently observed gender differences in affective and anxiety disorders are well established by the transition to adulthood. Females exhibit approximately double the rates observed for males in both the lifetime and past-year data. This apparent female disadvantage is somewhat lessened when attention deficit and hyperactivity disorders are added (38.6% for females vs 27.9% for males), and vanishes altogether when conduct disorder and antisocial personality disorder are also considered(46.6% in females vs 45.7% in males). This balance derives from the fact that boys are 1.75 times more likely to meet criteria for childhood conduct disorder and more than 2.5 times more likely to qualify for the antisocial personality diagnosis. In this latter category, 71% of the females and more than 59% of the males indicated that the behaviors involved were caused by their use of drugs or alcohol. Thus, not only are females much less likely to meet antisocial personality diagnostic criteria, those who do are more likely to attribute their behavior to substance use problems. When drug and alcohol disorders are also included, the prevalence of all study disorders combined is higher for males than for females (62.9% vs 58.5%, respectively). However, this difference is not statistically significant.
In general, 12-month prevalence rates correspond rather closely with lifetime rates, as might be expected given the youth of the cohort studied(Table 2). An important point in connection with these data is the tendency for disorders to be highly recurrent or persistent. For example, our analyses (not shown) reveal that only 22 % of 12-month prevalent cases of major depression turn out to be first onsets. When these cases are excluded from both the lifetime and 12-month prevalence estimates, the data indicate that 61% of those with a history of major depression experienced a recurrent or continuing episode of that disorder in the preceding year. Even in the case of alcohol dependence and marijuana dependence, in which 59% and 47%, respectively, of the 12-month prevalent cases were new onsets, approximately half of those with a history of alcohol or marijuana dependence had a recurrent or continuing episode in the year prior to interview.
Lifetime and 12-month prevalence rates by ethnicity and nativity are presented in Table 3 and Table 4. In examining these contrasts, we focus primarily on differences between African American and non-Hispanic white participants and between US-born and foreign-born participants within the Cuban and "other Hispanic" subgroups. Although when Axis I lifetime psychiatric disorders are considered together ("Any psychiatric disorder 1" in Table 3) no statistically significant differences are observed across ethnicity or nativity, many notable variations are evident for specific diagnoses. Compared with non-Hispanic white participants, African Americans are at a substantially lower risk for all study disorders considered together and for depressive disorder. They are also at dramatically lower risk with respect to the abuse of or dependence on substances, regardless of which substance-abuse or dependence category is considered. Among our African American sample, elevated lifetime prevalence is observed only with respect to PTSD. A similarly high occurrence of PTSD is found only among US-born "other Hispanics."
Comparisons across nativity among Cuban respondents revealed only one significant difference. Within this subgroup, the foreign-born respondents reported significantly higher levels of hyperactivity disorder than their US-born counterparts. In contrast, nativity is associated with a range of prevalence differences in the "other Hispanic" group. Higher rates were found among the US-born Hispanics on substance use disorders, conduct disorder, ADHD, and PTSD.
Because of interest in the possible significance of nativity as a gross index of acculturation, we also compared prevalence rates observed for US-born Cubans and US-born "other Hispanics" with those for foreign-born respondents of the opposite group. Statistically significant contrasts are indicated by superscript letters attached where relevant to each disorder listed. Except for PTSD, all of the differences found involved disorders characterized by externalizing symptoms. Importantly, in every instance, US-born respondents were shown to be at greater risk than those in the foreign-born comparison group. The 12-month prevalence data presented in Table 4 present variations across ethnicity and nativity that are substantially in accord with those found for lifetime prevalence.
Table 5 presents the prevalence of comorbidity in this community population, where comorbidity is defined as qualifying for 2 or more diagnoses throughout one's lifetime. The middle and lower portions of Table 5report this form of comorbidity when psychiatric and substance use disorders are considered separately. As shown in the first column of the top portion of the table, 39% of respondents had never experienced any of the disorders assessed, about 25% had experienced only 1 disorder, 12% had 2 disorders, 10% had 3, and more than 13% qualified for 4 or more diagnoses. From the second column, it can be seen that 42% of all lifetime disorders occurred in individuals whose histories included only that single disorder. In other analyses (not shown) we found that 62% of 12-month disorders were comorbid.
The patterns of comorbidity that are revealed when only psychiatric disorders are considered (middle portion of Table 5) correspond quite closely with those for all disorders. However, differences appear in analyses restricted to lifetime substance disorders. Not only are African Americans dramatically less likely to meet criteria for substance abuse or dependence (Table 3), those who do are significantly more likely than others to qualify for only a single substance use disorder. Somewhat lower rates of substance use disorder comorbidity are also found among foreign-born "other Hispanics," who, like the African American group, are comparatively unlikely to have experienced3 or more substance use disorders.
Comorbidity across substance and psychiatric disorders was also specifically examined. A total of 154 individuals (8.5% of the sample) met criteria for at least 1 substance use disorder and 1 psychiatric disorder. In 52.4% of these instances, the psychiatric disorder was reported to have occurred first; in 23.5% the substance disorder was first; and in 24%, the onsets of both types of disorder took place in the same year. It thus seems that psychiatric disorder is a better predictor of a subsequent substance disorder than is a substance use disorder a predictor of a subsequent psychiatric disorder.
Despite the youth of this cohort, or perhaps because of it, more than60% met criteria for 1 or more study disorders during their lives. In the vast majority of instances, the first onset of disorder occurred during the middle school and early high school years, and in 58% of the cases, detected disorders were comorbid. These findings indicate that there is a substantial presence of psychiatric and substance use disorders in middle and high school classrooms in South Florida. Thus, it cannot be assumed that study or intervention participants have no history of having a disorder just because they are young. This is a point that may not be well understood by many researchers and service providers.
The substantial level of comorbidity observed here falls well short of that reported by the NCS that involved participants up to 55 years of age. This age-associated contrast provides support for the argument that efforts might be usefully directed toward the primary prevention of secondary disorders.29 Because psychiatric disorders are more likely to precede the onset of substance use disorders than the reverse, it can be argued that efforts to prevent the occurrence of drug and alcohol problems should focus substantial attention on young persons who are experiencing or who have experienced a psychiatric disorder. Additional analyses (not shown) made clear that this increased risk for substance abuse or dependence applies to those with prior episodes of anxiety and/or affective disorders, whether or not the co-occurrences of conduct, antisocial personality, attention-deficit, and/or hyperactivity disorders are controlled.
Although expected gender differences in depressive and anxiety disorders are clearly confirmed in this young adult sample, gender equivalence in total prevalence was observed without the inclusion of substance diagnoses ("Any psychiatric disorder 2" in Table 1). This rather unique finding results from higher rates of hyperactivity disorder, ADHD, and conduct disorder among males—diagnoses that have not uniformly been assessed in prior investigations. When substance use disorders were also considered, the prevalence of "any study disorder" (Table 1) was 62.9% among males compared with 58.5% among females. The idea that women are at greater risk for mental disorders is not supported in these results, whether or not substance use disorders are taken into account.
The presence of Hispanics of differing nativity in this study allowed at least a gross estimation of the significance of acculturation as a risk or protective factor with respect to psychiatric and substance use disorders. Results obtained from this same cohort during the early adolescent years revealed lower rates of substance use among foreign-born compared with US-born participants,8,30 as well as better mental health.31 Similarly, Vega et al9 found lower rates of psychiatric disorders among Mexican immigrants than among their US-born counterparts. The findings in this study, with respect to the"other Hispanic" group but not within the subsample of Cuban heritage, substantially concur with these prior reports. While a lower prevalence was found among foreign-born "other Hispanics" for several psychiatric diagnoses, the more marked differences occurred for the substance use disorders. These latter contrasts were also observed in comparisons of this group with US-born Cuban immigrants.
Despite substantial research examining these differences,9,32 compelling explanations for why greater time spent in the United States is associated with increased mental health and substance abuse risk8,10 remain elusive. In this connection, it should be noted that none of the foreign-born participants in this cohort are recent immigrants, having come to the United States either before or at the time of entry into middle school. Thus, differences with their US-born counterparts seem to be persistent in nature. Evidence bearing on these issues will be presented in subsequent publications.
When psychiatric disorders other than conduct and antisocial personality disorders were considered together, no significant ethnic differences emerged; however, important differences in prevalence were observed for individual diagnoses and when all study disorders were considered together. Consistent with results from the NCS,2 African Americans have significantly lower rates of affective disorders, substance use disorders, and overall lifetime comorbidity than non-Hispanic white participants. However, contrary to the same study's report of no instances in which either lifetime or active prevalence was significantly higher among African American than white participants, we found significantly higher lifetime and 12-month prevalence of PTSD in the African American subgroup. This finding seems to be in accord with the results of Breslau et al.33
The limitations of this investigation include those that characterize prior studies that have derived diagnoses from a single structured interview that did not involve clinical judgment. Since the data are cross-sectional, lifetime prevalence estimates rely entirely on retrospective recall. While the young age of this cohort presumably minimizes recall problems, such problems remain a concern. Collectively, these measurement concerns require that the prevalence results reported be viewed as only estimates of the rates at which symptomatic experiences matching diagnostic criteria occur within the populations studied.
Accepted for publication June 26, 2001.
This study was supported by grant R01 DA10772 from the National Institute on Drug Abuse, Bethesda, Md.
Corresponding author: R. Jay Turner, PhD, Life Course and Health Research Center, Florida International University, University Park Campus, Deuxieme Maison (DM 243), Miami, FL 33199 (e-mail: email@example.com).