Racial and Ethnic Differences in Use of State-Operated Inpatient Substance Abuse Services, 2004–2005 Versus 2010–2011

This study was approved by the Yale University Human Investigation Committee and the CT DMHAS Institutional Review Board. Data extracts for these analyses were provided by the Information Services and the Evaluation, Quality Management, and Improvement Divisions of CT DMHAS from its Electronic Data Warehouse. At time 1, a random-extract method (that is, a computer-generated selection in which all unduplicated persons in each ethnic and racial category had an equal chance of being selected) was used to obtain a sample that was representative of the adult population (ages ≥18) that received substance abuse services at state-operated inpatient facilities in Connecticut from 2004 to 2005 (495 African Americans, 492 Hispanics, and 497 whites). For time 2, all persons identified as being African American (N=759; race coded as “black” and ethnicity as “non-Hispanic”), Hispanic (N=781; race coded as “white” or “other” and ethnicity as “Hispanic”), and white (N=2,989; race coded as “white/Caucasian” and ethnicity as “non-Hispanic”) in fiscal year 2010–2011 were included in the analysis. Clients who received services at different types of inpatient addiction treatment centers (e.g., detox, short-term inpatient, and long-term inpatient) were included in the analyses. Persons with missing race or ethnicity information and groups with limited representation (that is, clients from a multiracial background or with multiple ethnicities or clients from other races) were excluded.

To assess the relationship between race-ethnicity and treatment-related variables, logistic and linear regressions were conducted, controlling for demographic factors and symptom severity. The controlling variables (Table 1) were entered first, followed by race-ethnicity. The same data analytic procedures were conducted for times 1 and 2. ORs <1 indicate lower odds, and ORs ≥1 indicate higher odds. ORs can also be interpreted as measures of effect size in order to quantify the magnitude of change over time. For length of stay, we conducted linear regressions, reporting standardized betas.

The dependent variables included referral source (self-referral, inpatient services, criminal justice system, and crisis or emergency services), legal status at admission (voluntary, criminal justice system, and emergency certification), discharge reason (facility concurs or does not concur with discharge), DSM-IV-TR axis I diagnosis at admission and discharge (alcohol and drug use disorders,) and axis II diagnosis (personality disorders cluster A, B, and C, personality disorders not otherwise specified [NOS], intellectual disability or borderline IQ, and diagnosis deferred), and length of stay (number of days). Cluster B personality disorders are histrionic, narcissistic, borderline, and antisocial. Categorical variables were recoded as yes or no.

Table 1 shows the results of the logistic and linear regressions at times 1 and 2 (only findings that were significant at time 1 or time 2 are shown). [Tables presenting significant results of the full models from the regression analyses and summarizing differences in demographic characteristics are included in an online supplement to this report.] At time 1, Hispanics were more likely than whites and African Americans to self-refer to services. In contrast, at time 2, Hispanics were less likely than whites to self-refer. At both time points, African Americans were more likely than whites and Hispanics to be referred by the criminal justice system, although the ORs at time 2 were smaller than at time 1, suggesting smaller effect sizes. Furthermore, at both time points, Hispanics were more likely than whites to be referred by other sources (for example, family and friends), with a smaller effect size at time 2.

With respect to legal status, at times 1 and 2, Hispanics were more likely than whites to begin treatment voluntarily, with a larger effect size at time 2. At time 1, African Americans and Hispanics were less likely than whites to be committed to treatment on an emergency basis. At time 2, only the African American–white difference remained significant, with a larger effect size than at time 1.

At times 1 and 2, length of stay for African Americans was longer than that of the other groups. In contrast, length of stay for Hispanics at both time points was shorter than for whites and African Americans.

At time 1, whites were more likely than the other groups to be diagnosed as having an alcohol use disorder. At time 2, African Americans were more likely than whites to be diagnosed as having an alcohol use disorder. At both time points, Hispanics were the group least likely to be diagnosed as having an alcohol use disorder. At time 1, Hispanics were more likely than whites or African Americans to receive diagnoses of drug use disorders, and African Americans were more likely than whites to receive these diagnoses. At time 2, Hispanics continued to be the group most likely to receive a diagnosis of a drug use disorder, and African Americans were less likely than whites to receive a diagnosis of a drug use disorder.

Hispanics were more likely than whites to receive a cluster B personality disorder diagnosis at time 2 admission and at times 1 and 2 discharge. At time 1 discharge, African Americans were more likely than whites to receive this diagnosis. In addition, African Americans were more likely than Hispanics to receive a diagnosis of personality disorder NOS at time 1 admission, and Hispanics were more likely than whites to receive that diagnosis at time 2 admission.

At time 1 admission, Hispanics were more likely than whites to receive a deferred axis II diagnosis; however, at time 2, whites were more likely than Hispanics to receive this diagnosis. At time 2 admission, African Americans were less likely than whites to receive a deferred axis II diagnosis. At time 2 discharge, Hispanics were also less likely than African Americans to receive this diagnosis.

These analyses compared racial-ethnic differences in state-operated substance abuse inpatient treatment units in Connecticut in two time periods: 2004–2005 and 2010–2011. Differences in self-referral changed between time points. At time 2, Hispanics were less likely to self-refer than whites, which is consistent with previous findings (11). Self-referrals imply an internal motivation to seek services; however, self-referrals may also reflect external motivations (for example, encouragement from one’s social network) (12). Racial-ethnic differences in referral by the criminal justice system remained significant between time points, with African Americans being more likely than other groups to be referred by this source. This finding is consistent with extant literature (1,11,13). Furthermore, the direction of differences in referral from other sources did not change between times 1 and 2. Congruent with previous studies, Hispanics were more likely than whites to be referred by other sources.

Differences in length of stay also remained significant between time points. African Americans tended to stay longer in treatment than whites or Hispanics, which may be the result of compulsory requirements from the criminal justice system or biases regarding this racial group. Studies have suggested that mandated as opposed to voluntary treatment is associated with less attrition and better outcomes (1,13). In contrast, and in line with other studies (14), length of stay for Hispanics was shorter than for whites, which may be attributable to a scarcity of bilingual-bicultural staff.

The findings at time 1 that whites were more likely than the other groups to have a diagnosis of an alcohol use disorder and that African Americans were more likely than the other groups to have a diagnosis of a drug use disorder are in line with previous findings (3,4), but the differences found at time 2 are not supported by the literature, warranting further investigation.

Previous research suggests that Hispanics are less likely than whites to be diagnosed as having personality disorders (15), which was not supported by our findings. However, when comorbid substance use and personality disorder diagnoses were investigated, Hispanics were more likely than whites to receive diagnoses of both personality disorders and alcohol use disorders and more likely than African Americans to receive diagnoses of both personality disorders and drug use disorders. (15). Thus a combination of substance use disorders and personality disorders might explain our findings at time 2.

Even with the large sample and the availability of data from two periods, this study had several limitations. First, the data were provided by a behavioral health system in a single state, which limits the generalizability of findings. Second, the data were not longitudinal; the set of participants studied at time 1 did not match the set at time 2, limiting the ability to track participants included in both samples and examine change over time. Third, the differences found might be due to secular factors, such as demographic characteristics, changes in practice patterns, and nonstandardized diagnostic methods.

In addition, our race-ethnicity categorization might not reflect the heterogeneity of each group. Hispanics, for instance, constitute a diverse ethnic group that varies in race. Excluding individuals identified as both black and Hispanic (which we did for consistency between times 1 and 2 analyses) is contrary to current analytic strategies. Finally, our study was exploratory in nature. Future investigations might benefit from testing specific conceptual frameworks.

Despite its limitations, this study had important strengths. To our knowledge, this is one of only a few investigations examining addiction-related differences across significantly large ethnic-racial samples at independent time points. Moreover, this study examined the relationship between multiple treatment-related variables and diagnosis and race-ethnicity.

This study showed that many of the racial-ethnic differences documented in 2004–2005 remained in 2010–2011, with the exception of self-referral. The findings highlight the importance of identifying systemic issues and continuing to evaluate culturally competent practices aimed at addressing inequities as well as investigating the association between treatment-related variables, diagnosis, and treatment outcomes. Closer examination of the specific findings may help determine targets for intervention. For instance, the diagnostic differences found in this study are in contrast with data about the base rates of these disorders in the community. Are the differences the result of diagnostic bias or diagnostic practices that need to be addressed? Similarly, understanding racial-ethnic differences in treatment attrition and completion is important, and reasons for these differences, such as referrals from the justice system for African Americans, should be investigated.