The major finding of this study is that service utilization patterns among black and white respondents depended on the specific disorder and type of treatment considered. Black respondents were consistently less likely than white respondents to receive services for mood and anxiety disorders, even after we adjusted for predispositional, enabling, and need factors associated with treatment utilization. This was true for the full sample, persons with any axis I disorder, and those whose need was specific—in other words, a mood or anxiety disorder—confirming prior findings that the black population receives fewer mental health services than the white population for treatment of mood and anxiety disorders (1
In contrast no evidence emerged of service utilization disparity between black and white respondents for alcohol or drug use disorders, despite statistical power to detect a difference. Black respondents in the full sample and among respondents with a drug use disorder were significantly more likely than white respondents to receive drug services once differences between the groups on predispositional, enabling, and need factors were controlled. Some general population studies on race differences in substance treatment indicated no differences (3
), whereas others have suggested that the black population is less likely to utilize mental health services for substance use disorders (17
); however, these studies lacked specific diagnostic information on a full range of adults in a national sample. It should be noted that less than 20% of people with a drug or alcohol use disorder used any service, which highlights the disparity of service utilization across racial-ethnic lines.
The contrast between lower service utilization rates for black respondents with mood and anxiety disorders and equal to or higher rates than whites with alcohol use or drug use disorders raises important questions about differential service utilization patterns. Although several past studies showed that racial-ethnic differences in mental health service utilization are substantially attenuated by socioeconomic status differences (7
), this investigation indicated that disparities persisted despite differences in income or education. Other explanations may include cultural and policy-level factors. Culturally, black communities may react more negatively than white communities to hazardous drinking (61
) and provide greater social support for sobriety (62
). Cultural influences may affect conceptions of causes and treatments of mental health problems, including perceived need for treatment (11
), resulting in differences between black and white communities (63
). Medical research has consistently shown that black persons are more likely to distrust the medical community and the health care system (16
). There may also be important subgroup differences within the black community; for example, recent evidence indicates that African Americans may utilize more mental health services than Caribbean blacks (65
). Future studies using more refined racial-ethnic categories could further specify mediators of racial-ethnic differences, including nativity and acculturation, which may affect perception of the mental health care system. Racial-ethnic attitudes toward anxiety or depression have not been directly compared with attitudes toward alcohol or drug use disorders, an important direction for future research.
Furthermore, questions about the reasons for the different patterns of services in black and white populations could focus on potential confounding due to specific subgroups within the general samples we examined. This is addressed in part elsewhere (59
) by examining treatment disparities among the subgroup of whites and blacks that had mood or anxiety disorder co-occurring with an alcohol or drug use disorder. In this subgroup analysis, we also found significant differences: white respondents were 2.05 times as likely as black respondents to use mood or anxiety services during their lifetime and .80 times as likely to use drug services, and again we found no difference in using alcohol services (OR=.99). Thus patterns of racial-ethnic differences in the comorbid subgroup were generally similar to the patterns shown above, despite the potential greater severity of illness among those with more than one type of disorder. This is important information because comorbidity is a strong predictor of treatment entry. Further, we examined whether the patterns of axis I treatment utilization were different among the subset of respondents with an axis II diagnosis. Results were generally similar, although there were no significant racial-ethnic differences in drug treatment service utilization among individuals with a drug use disorder. Taken together, the results of both investigations indicate that ethnic differences in treatment utilization are not substantially attenuated or modified by comorbidity, and furthermore, the magnitude of difference is remarkably similar in the comorbid subgroups compared with the broader group of all individuals who received a diagnosis as analyzed in this study.
At a policy level social coercion through law enforcement—an “external environment” enabling characteristic of the Andersen model (23
)—might indirectly explain higher service utilization for substance use disorders among black individuals because drug policies and laws disproportionately affect black communities (66
). At similar toxicology levels, more persons in minority groups than whites are mandated to receive alcohol treatment by the criminal justice system (68
). Also, some evidence indicates that black persons who enter the mental health system are more likely than whites to have entered through coercion (70
). Future studies should determine how social pressure and legal coercion influence entry into services for different disorders and how this affects larger perceptions of treatment for the different disorders by race and ethnicity. In addition, mental health service delivery systems have changed over time; determining that disparities vary across changes in the mental health service system over time would suggest mechanisms through which disparities exist.
We now address limitations of the study. First, in the examination of lifetime disorders and service utilization, respondents' recall bias might have affected estimates, particularly among older respondents. To investigate this, we reran our analysis limited to current (past 12 months) disorders. The magnitude and direction of effects were virtually unchanged, suggesting that recall bias did not substantially affect our estimates. Smaller numbers of respondents in diagnostic categories limited the power to detect effects, so for this reason, as well as the high quality of the lifetime diagnoses provided by the AUDADIS-IV (35
), we reported lifetime measures. Recall bias can also affect the type and timing of treatment reported. We assume that there would be no racial difference in recall, but evidence (which is lacking) on the accuracy of this assumption would assist in interpreting the results.
Second, individuals institutionalized for mental disorders or incarcerated throughout the data collection were not included; disparities in institutionalization could bias the findings. Persons from the black community have higher incarceration rates, but studies are inconsistent on whether persons belonging to minority groups receive fewer mental health services in prison than whites do (74
). Racial-ethnic differences in rates of long-term mental health institutionalization are also unclear (15
), so institutionalization effects on our findings are uncertain. Third, these data were cross-sectional. Longitudinal data would allow prospective examination of factors predicting entry into mental health services, which will be possible when NESARC follow-up data become available.
The NESARC did not focus solely on treatment utilization; additional detail on treatment in future surveys would be helpful. However, differences in questions regarding mood and anxiety treatment and alcohol or drug treatment utilization reflect true differences in both the services and service delivery systems for these disorders. In addition, racial-ethnic differences reflect differences in services available as well as individual treatment choices. Future surveys should include information on services available to the respondent. Finally, this study addressed disparities in service utilization but could not address other important disparities, such as quality, appropriateness, and efficacy of care. Although disparities in quality of care within treatment services have been shown (68
), this study highlights disparities in receiving any care at all.