Race-Ethnicity and Diagnosis as Predictors of Outpatient Service Use Among Treatment Initiators

Mental health and substance abuse problems are among the most prevalent and costly disorders in the U.S. health care system, and they often co-occur ( 1 ). Most people with mental disorders, including substance use disorder, do not receive appropriate quality of care ( 2 ). Racial and ethnic minority status is associated with greater mental health and substance abuse symptom severity and frequency ( 3 ). Additionally, persons from racial or ethnic minority groups are less likely than non-Latino whites to use mental health care ( 4 , 5 ). Recent research from the National Comorbidity Survey Replication indicates that compared with non-Latino whites, non-Latino blacks and Latinos are less likely to receive treatment for mental illness and substance use disorders over the previous 12 months and have longer delays in receiving treatment for at least some disorders ( 6 , 7 ). Significantly lower utilization rates of mental health care have been found among blacks and Latinos, compared with non-Latino whites ( 8 , 9 ). From 52% to 73% of people with substance use disorders eventually make treatment contact; delay in seeking treatment is associated with many sociodemographic characteristics, including race and ethnicity ( 7 ). Income, education, and ethnicity significantly affect the probability of having a regular relationship with a mental health specialist ( 10 ).

Epidemiological studies examining racial-ethnic disparities in mental health treatment often include people who are currently not in treatment ( 6 , 7 ). Our study addresses this gap in the literature by focusing on care for mental and substance use disorders among individuals who have already initiated contact with specialty mental health services in order to examine factors associated with continuity of care. We aimed to examine the relationship between race and ethnicity, a diagnosis of mental illness and substance use disorder, and baseline functioning and the extent of care for mental or substance use disorders received after an initial intake at an outpatient specialty mental health or substance abuse treatment center.

Building on previous research, we had three hypotheses. First, we hypothesized that after the analysis controlled for age, sex, education, and diagnostic group, non-Latino blacks and Latinos would report worse mental health and substance abuse functioning than non-Latino whites. Second, we hypothesized that after the analysis controlled for client characteristics stated above, individuals with dual diagnoses of mental and substance use disorders would receive more outpatient visits than those with a diagnosis of a mental disorder only or a substance use disorder only. Third, we hypothesized that after the analysis controlled for the client characteristics stated above, non-Latino blacks and Latinos would have had fewer outpatient visits than non-Latino whites.

Methods

Sample

The sample consisted of 1,899 consecutive new outpatient intakes (defined as an intake for patients not seen in the previous six months) for English-speaking adults. Patients in the sample arrived for treatment at one of 12 participating sites during the study period (May 2001 through June 2002), had information available from administrative data sources about the number of outpatient visits in the two months after the intake, and had a race-ethnicity that could be categorized as non-Latino white, non-Latino black, or Latino. Participating sites were outpatient mental health, dual diagnosis, or substance abuse programs. Day or partial hospitalization programs were not included. Sites were instructed to include all consecutive intakes during the study period. However, because some sites were unable to provide us with the total number of intakes during the study period, we were unable to determine the participation rate. Mental health and substance abuse treatment centers in the four major geographic U.S. census regions were included in the study and consisted of private and public institutions representing a variety of outpatient settings.

Procedure

Symptom surveys for mental and substance use disorders and demographic questions were administered twice: upon intake and four to eight weeks after intake. Because this study focused only on symptoms of mental and substance use disorders at intake, follow-up scores for this group, which have been previously reported in the literature ( 11 ), are not reported here. Surveys were administered by program staff within the context of continuous quality improvement programs as part of routine outcomes monitoring. Verbal consent was obtained from all participants. Participants responded to the survey on mental and substance use disorders in the waiting room of the delivery site at the intake appointment, and they returned the survey to the clinic staff. Admission and discharge dates, DSM-IV psychiatric diagnoses, and the number of visits in the two months after intake were extracted from medical records or administrative databases. All surveys and corresponding administrative data were sent to the principal investigator. This data collection process was approved by the institutional review board of McLean Hospital and Boston University Medical Center and by each participating site.

Measures

Symptoms of mental and substance use disorders. The 24-item Behavior and Symptom Identification Scale (BASIS-24) is a self-report measure that assesses treatment outcomes for mental and substance use disorders. The BASIS-24 is a brief, responsive, reliable, and valid measure designed to assess outcomes of mental health treatment from the consumer's point of view ( 11 ), and it has been validated among the three largest racial groups in the United States: non-Latino whites, non-Latino blacks, and Latinos ( 3 ).

Six domains, 24 items in total, comprise the BASIS-24: depression and functioning (six items), interpersonal relationships (five items), self-harm (two items), emotional lability (three items), psychosis (four items), and substance use disorders (four items). Each item asks a question about how the respondent is feeling in different areas of life according to five ordered response options reporting either the level of difficulty experienced (no difficulty to extreme difficulty) or the frequency with which a symptom or problem has occurred (none of the time to all of the time). Examples of questions include "During the past week, how much difficulty did you have coping with problems in your life?" (depression and functioning); "During the past week, how often did you have an urge to drink alcohol or take street drugs?" (substance abuse); and "During the past week, how often did you hear voices or see things?" (psychosis). The complete instrument and information about how to obtain it are available at www.basissurvey.org.

Item response theory (IRT) analyses were used to create standardized scores for each domain of the instrument and to compute an overall summary score ( 11 , 12 ). However, because computation of IRT scores requires specialized software not available to many mental health programs, a linear approximation to the IRT scores was developed by performing an ordinary least-squares regression of the IRT domain and overall summary scores by using raw scores as independent variables. The coefficients were used as weights to compute weighted scores that maintain the same range of values (0–4) as with the BASIS-24 items. The weighted scores have correlations with the IRT scores ranging from .96 to >.99. Lower scores indicate less symptom or problem difficulty or frequency, and higher scores indicate greater symptom or problem difficulty or frequency ( 11 ).

Diagnoses. Primary and secondary DSM-IV diagnoses were collected from medical records or administrative databases. For analysis purposes, clients who had a primary or secondary (if applicable) axis I mental illness diagnosis that did not involve alcohol or drug use disorders were classified in a mental disorder-only diagnosis category (N=1,058, or 56%). Clients with an axis I alcohol or drug use disorder as either a primary or a secondary diagnosis (if applicable) were classified in a substance use disorder-only category (N=453, or 24%). Clients with a mental illness and an alcohol or drug use disorder as both primary and secondary diagnoses were classified in a category of mental and substance use disorders (N=296, or 16%). The remaining 5% (N=92) of clients either did not have an axis I diagnosis or diagnostic information was not available from the records, and they were not included in the analyses. In addition to these diagnostic categories, the following primary diagnoses were created: schizophrenia, schizoaffective, and other nonaffective psychotic disorders; depressive disorders; bipolar disorder with manic or mixed episodes; and alcohol or drug use disorders.

Outpatient visits. The number of outpatient visits in the two-month period after the initial intake at a mental health or substance abuse treatment center was extracted from medical records or administrative databases.

Analyses

Using multiple regression, we examined the effects of race-ethnicity and diagnoses on functioning at intake in regard to mental and substance use disorders (hypothesis 1) and number of outpatient visits (hypotheses 2 and 3). In testing hypothesis 1, a separate regression analysis was run for each of the six BASIS-24 domains, and one regression analysis was run for the overall score, with the predictors of race-ethnicity added into the model, along with the covariates of age, sex, education, and diagnosis. Multicollinearity was assessed by using Pearson correlation coefficients.

To test hypotheses 2 and 3, a regression analysis was run with predictors of race-ethnicity and diagnosis category (mental disorder only, substance use disorder only, and mental and substance use disorders) added into the model, along with covariates of age, sex, education, diagnosis, and severity of baseline symptoms as measured by the BASIS-24 overall score. The BASIS-24 subscale scores are highly correlated with the overall score; therefore, only the overall score was used as a covariate in the model. Because of multicollinearity between the diagnostic category of substance use disorder only, a category based on primary and secondary diagnoses, and the client's medical record diagnosis of alcohol or drug use disorder (Pearson r=.90), the diagnosis of alcohol or drug use disorder was removed from the analysis for hypotheses 2 and 3. Although we did not have a specific hypothesis regarding differences in mental health functioning and number of outpatient visits between Latinos and non-Latino blacks, we examined these differences by replicating the regression analyses by using Latinos rather than non-Latino whites as the reference group. All analyses were carried out with SPSS, version 15.

Imputation of missing data

Although missing data rates were low (ranging from <1% to 3.3% across all items), we imputed missing BASIS-24 ratings and assumed data were missing at random, conditional on age and gender. The missing at-random assumption implies that the probability that an observation is missing depends on all the observed data but not on the unobserved value of the missing data. We used the SAS procedure PROC MI to generate values by modeling the nonmissing items, gender, and age from a multivariate normal distribution ( 13 ). For each missing value we generated an observation by using a Markov Chain Monte Carlo algorithm based on starting values from the expectation-maximization algorithm. Because item responses are ordinal, we rounded the imputed value to the nearest integer ( 11 ).

Results

Client and site characteristics are presented in Table 1 for the total sample, with demographic differences noted between the three racial-ethnic groups. We present univariate data by race-ethnicity and by BASIS-24 domain scores. We also present stratified scores by diagnostic category because BASIS-24 domain scores are correlated with primary diagnosis and because distribution of diagnoses varied among the three racial-ethnic groups ( Table 2 ). Significant differences in mean scores among all three racial-ethnic groups were found for each of the BASIS-24 domains and for the overall score. Among participants with schizophrenia or schizoaffective disorder, Latinos reported greater emotional lability, compared with non-Latino whites and non-Latino blacks. Among participants with depressive disorders, Latinos and non-Latino blacks reported worse interpersonal functioning, more severe psychosis symptoms, and worse overall functioning than non-Latino whites; however, Latinos alone reported more severe self-harm symptoms. Moreover, Latinos with alcohol or drug use disorders reported worse interpersonal functioning, compared with non-Latino whites and non-Latino blacks. Additionally, non-Latino blacks with alcohol or drug use disorders reported greater self-harm symptoms than non-Latino whites with such disorders.

Enlarge table

Enlarge table

Relationship between race-ethnicity and mental health functioning

Hypothesis 1 stated that non-Latino blacks and Latinos would report worse functioning because of mental and substance use disorders, compared with non-Latino whites. We controlled for the covariates of age, sex, education, and diagnoses, because these were significant predictors of mental health functioning and the purpose of this hypothesis was to explore the effects of race-ethnicity on mental health functioning. This hypothesis was upheld, in that in almost all BASIS-24 domains non-Latino blacks and Latinos reported greater symptom and problem frequency or severity than non-Latino whites, even after the analyses controlled for patient characteristics in the regression model ( Table 3 ) (F=20.78, df=15 and 1,883, p<.001; adjusted R ² =.14). The exception to this was in the depression and functioning domain, where after the analysis controlled for the covariates, symptoms were similar among the three racial-ethnic groups. The only significant difference between Latinos and non-Latino blacks was that non-Latino blacks showed greater severity of substance use disorders than Latinos ( β =.12, p<.001, data not shown).

Enlarge table

Race-ethnicity, mental health diagnoses, and mental health visits

Hypothesis 2 stated that after the analysis adjusted for client characteristics, clients with dual diagnoses of mental and substance use disorders would have more outpatient visits than those with a diagnosis of a mental disorder only or a substance use disorder only. Hypothesis 3 stated that after the analysis adjusted for client characteristics, non-Latino blacks and Latinos would have fewer outpatient visits than non-Latino whites. These hypotheses were tested in one regression equation with the number of outpatient visits as the dependent variable ( Table 3 ). Although unadjusted data showed that non-Latino blacks appeared to have more outpatient visits than non-Latino whites ( Table 4 ), race-ethnicity had no significant effect on the number of outpatient visits after the analysis adjusted for the other demographic and diagnosis variables. Similarly, when the analyses used Latinos as the reference group, there was no difference between Latinos and non-Latino blacks in the number of outpatient visits (data not shown). A diagnostic category of substance use disorder only or a category of dual diagnosis, as well as severity of overall mental health functioning, were significant predictors of a greater number of outpatient visits, above any effect of client race-ethnicity, a specific axis I diagnosis, age, education, or gender (F=12.18, df=17 and 1,783; p<.001, adjusted R ² =.10). R ² change statistics indicated that the best model fit involved regressing client race-ethnicity and diagnostic category onto number of outpatient visits (change statistics: R ² =.10, F=76.23, df=2 and 1,786, p<.001). Inclusion of diagnosis, age, education, and sex as covariates in the model increased the variance explained by less than 1% (p=.07).

Enlarge table

Discussion

This study aimed to examine the relationship between race-ethnicity, diagnoses, and mental health functioning and the number of outpatient mental health and substance abuse treatment visits among individuals who had already initiated contact with an outpatient mental health or substance abuse treatment center. Evidence was found to support the hypothesis that mental health functioning is worse among Latinos and non-Latino blacks who have initiated treatment, compared with non-Latino whites. This finding is contrary to recent evidence reporting lower rates of mental health problems among non-Latino black and Latino populations in treatment for substance abuse ( 14 ). Data from the study presented here suggest that after the analysis controlled for client characteristics, non-Latino blacks and Latinos reported worse mental health functioning in five of the six domains measured by the BASIS-24: interpersonal functioning, self-harm, emotional lability, psychosis, substance use disorder, and overall mental health. Non-Latino whites, non-Latino blacks, and Latinos reported similar depressive symptoms and functioning.

Despite this difference in mental health functioning among these two racial-ethnic minority groups and whites, our study did not provide any evidence for increased outpatient visits among these racial-ethnicity minority groups. Non-Latino white, non-Latino black, and Latino clients had a similar number of outpatient visits in the two months following intake after the analysis controlled for age, sex, education, and diagnosis. Substance use disorders alone and in combination with other mental health conditions and severity of overall baseline functioning were significant predictors of the number of outpatient visits, whereas race-ethnicity was not. Again, these data do not support earlier findings suggesting that among those with a dual disorder, non-Latino whites are more likely than Latinos or non-Latino blacks to receive treatment for substance use and mental disorders ( 15 ).

Examining patients' perceptions of mental health and treatment among those who have successfully engaged in treatment may provide information on why clients from racial-ethnic minority groups have worse mental health functioning at intake yet similar rates of continuation in treatment, as was found in our study. Recent studies have found that a lack of prior helpful treatment experiences, a lack of confidence in currently practiced mental health treatment models ( 16 , 17 ), stigma ( 18 , 19 , 20 ), or past experiences with discrimination ( 21 ) either prevented adults from seeking treatment or were associated with delays in seeking mental health treatment. Increasing patient activation for mental health treatment is one way to ensure appropriate care ( 22 ). Patient activation encompasses being a collaborative partner in managing one's health and treatment, including accessing and navigating the health care system ( 23 ). A recent evaluation of a patient activation strategy in mental health care found that in an intervention that served primarily Latinos, participants were more likely to remain in treatment and to schedule and attend more appointments than those who did not receive the activation intervention ( 24 , 25 ).

Although our study was relatively large in scale, one of its limitations was that participants were taken from a convenience sample of patients with an outpatient intake at a mental health treatment center. These patients may not be representative of the U.S. population of individuals in treatment for mental or substance use disorder. Because we do not have data on the number of intakes during our study period, we cannot determine the response rate in our study. Moreover, we have little information on the specific content of each program, and therefore we cannot determine how program characteristics may affect number of visits. Additionally, clients' insurance status, which may be related to access to care, was not included as a predictor of service use.

Our study did not find racial or ethnic disparities in mental health or substance abuse service use among clients who had already initiated outpatient services. These clients may differ from those with mental or substance use disorders who have not sought or received treatment. This suggests that once clients overcome barriers to seeking and receiving treatment, service use may be equal across racial-ethnic groups. Our study found that non-Latino blacks and Latinos had worse mental health functioning than non-Latino whites, yet all three groups had similar utilization of specialty outpatient mental health care. After adjustment for symptom severity and diagnoses, we also found no differences in number of outpatient visits among the three racial-ethnic groups. Further research will need to examine two things: First, among people beginning treatment for mental and substance use disorders, how do specific components of mental health and substance abuse services lead to continued or discontinued care? And second, among people currently not in treatment, why might perceptions of mental health treatment and feelings of activation to initiate mental health treatment vary by race and ethnicity?

Acknowledgments and disclosures

This study was supported by grant R01-MH-58240 from the National Institute of Mental Health (principal investigator, Dr. Eisen) and in part by the Department of Veterans Affairs Health Services Research and Development program. The views expressed in this article are those of the authors and do not necessarily represent those of the Department of Veterans Affairs.

Dr. Eisen was the developer of the BASIS-24. The other authors report no competing interests.