The Role of Providers in Mental Health Services Offered to American-Indian Youths Arlene Rubin Stiffman, Ph.D. Stacey Freedenthal, Ph.D. Peter Dore, M.S. Emily Ostmann, M.P.H. Victoria Osborne, M.S.W. Hiie Silmere, M.S.W.
Objective: American-Indian adolescents have high rates of addiction and mental health problems but low rates of service use. The gap between service need and use appears to be even larger than the known gap for the general population, and few of the services are provided by specialists. This study examined receipt of treatment by AmericanIndian youths for addictions or mental health problems, the service provider who first identified a problem and sent a youth to treatment, and the extent to which the provider’s knowledge and assessment predicted variance in service actions. Methods: A sample of 401 AmericanIndian youths (196 from an urban area and 205 from a reservation) aged 12 to 19 years was first interviewed in person in 2001. A total of 188 of the youths’ treatment providers were then interviewed. Results: Structural equation modeling showed that 30 percent of the variance in addictions or mental health services provided to youths was predicted by the provider’s assessment of the youth’s mental health, the provider’s resource knowledge, and provider type. Conclusions: The results demonstrate that professional, informal, and traditional providers play a pivotal role in providing treatment services offered to American-Indian youths and that these providers were more likely to identify a youth’s problems and to offer and refer services when the provider knew more about community resources for the youth and about the youth’s personal and environmental problems. (Psychiatric Services 57:1185–1191, 2006)
A
merican-Indian adolescents have high rates of addiction and mental health problems (1) but low rates of service use (2). The gap between service need and use appears to be even larger than the known gap for the general population
(3,4), and few of the services are provided by specialists (5,6). This study examined the roles of providers in moderating the gap between need and services among American-Indian youths (7). In general, whether youths are of-
Dr. Stiffman, Mr. Dore, Ms. Osborne, and Ms. Silmere are affiliated with the George Warren Brown School of Social Work, Washington University, 1 Brookings Drive, Campus Box 1196, St. Louis, MO 63130 (e-mail:
[email protected]). Dr. Freedenthal is with the Graduate School of Social Work, University of Denver. Ms. Ostmann is with the St. Louis County Health Department.
PSYCHIATRIC SERVICES
♦ ps.psychiatryonline.org ♦ August 2006 Vol. 57 No. 8
fered or referred to services strongly depends on an adult’s awareness of problems and knowledge of service resources (7). When functioning in this role, these adults might be called “gateway” providers (8) because they open the gateway to services for youths. The providers may be professionals (mental health or addictions specialists or providers from primary health care, child welfare, juvenile justice, and education) (9) or informal providers (parents and respected elders) (10,11). In the American-Indian community, they might also be traditional providers (healers, medicine people, and ceremonial leaders) (5,12,13). Regardless of the category of gateway provider, recognition of youths’ service need is influenced by a number of factors. Gateway providers are more likely to identify a youth as needing services if the youth has functional impairments, comorbid conditions, and other risk factors (14). Similarly, they are more likely to recognize need if a youth has many problems or particularly visible problems (15). Training might improve identification because mental health professionals (for example, psychologists with doctorates and psychiatrists) identify a higher number of needy youths than other care professionals who help youths (16). The gateway provider model was developed by Stiffman, Pescosolido, and Cabassa (7) in 2001 and is shown 1185
Figure 1
Gateway provider model, demonstrating the moderating role of the provider between need and service access among youths Youth’s need and predisposing factors Youth’s mental health problems
Youth’s environment
Gateway provider characteristics
Services
Provider’s assessment of youth’s mental health and environment
Provider type and training
Mental health services offered
Provider’s resource knowledge
in Figure 1. It adds to prior service models (17,18) by showing the moderating role of the provider between need and service access. On the basis of this model, we hypothesized that providers’ assessment of youths’ mental health and environment (determined by provider reports) and providers’ resource knowledge (determined by provider training) predict services offered for AmericanIndian youths, as they do in other populations (19). To test this hypothesis, we used self-report data from youths and their providers that were collected during a study funded by the National Institute on Drug Abuse. The data included service use and drug use information for two American-Indian populations (urban and reservation) from a single southwestern state.
Methods Sample In 2001 a total of 401 youths (205 from a reservation and 196 from an urban area) aged 12 to 19 were interviewed for the American Indian Multisector Help Inquiry. The youths were randomly selected from complete tribal enrollment and school district records. Only one child per household was enrolled (5). The youths’ mean±SD age was 15.5±1.5 years, and 176 (44 percent) were male. 1186
Institutional review boards at Washington University, the tribal council, and the urban school district shaped consent and protection procedures (5). Personnel from local American-Indian educational and health services made the initial contact with the families. Only six families or youths refused in each area (12 total). Youths who had received help in the prior year (2000) for mental health, addictions, or behavioral problems identified the individuals who helped them. The youths said they were served by 235 informal providers, 225 professional providers, and 47 traditional providers, but 46 of the youths who had a provider (about 20 percent) said that they did not know the provider’s name. Within six months, 188 of those providers were interviewed. Seventy of the providers interviewed were informal helpers (38 were parents or foster parents, 27 were other relatives, and five were from another group of informal helpers), seven were traditional healers, and 111 were professionals (96 nonspecialists and 15 specialists). Although the youths typically named multiple informal helpers, we sought to interview only each youth’s primary informal helper and all professional (specialists and nonspecialists) and traditional proviPSYCHIATRIC SERVICES
ders. Provider refusal rates were less than 5 percent. We merged data on providers and youths to yield two data sets: one with 188 unique providers but repeated youths (more than one provider had served the same youth) and one with 141 unique provideryouth pairs. We present data only from the former in this article. Interview procedures In 2001 trained interviewers contacted the youths for whom guardian permission was granted and administered the interviews. The field supervisors and most interviewers were American Indian. Interviewers were required to accurately and smoothly complete a practice interview before entering the field. All interviews were audiotaped for monitoring and backup purposes. Supervisors, who gave immediate feedback, reviewed each interviewer’s entire first two interviews and selected sections of further interviews. Instruments For structural equation modeling (SEM), manifest variables were used to create latent variables (20). Table 1 shows the alphas, means, and factor loadings of latent constructs in order to determine their validity in SEM and the gateway provider model. Data on youths Youths’ report of mental health problems. A latent variable was created through factor analyses of symptom counts for questions concerning suicide, depression, posttraumatic stress disorder symptoms, conduct disorder, violence, and substance (alcohol and other drugs) abuse or dependence. The questions came from the National Institute of Mental Health’s Diagnostic Interview Schedule (21). Identifying the gateway providers. The Service Assessment for Children and Adolescents (22) was used to identify individuals who helped the youths. The modules were refined to conform to services available to American-Indian youths from the communities studied, including informal help and traditional AmericanIndian healing services.
♦ ps.psychiatryonline.org ♦ August 2006 Vol. 57 No. 8
Youths’ perceived environment. A latent variable was created from seven manifest variables concerning neighborhood characteristics (23): drug dealing, shootings, murders, abandoned buildings, neighbors receiving welfare, homeless people on the street, and prostitution. Each question was scored with a 0, none; 1, some; or 2, a lot. Data on providers Provider surveys included two sections: a general-approach module with questions about providers’ background and a youth-specific module with questions for providers about the specific youth and actions on behalf of the youth. A latent variable indicating the providers’ assessment of youths’ mental health problems was calculated from two multi-item manifest variables: the providers’ report of youths who presented with problems and the providers’ rating (on a 5-point scale, ranging from 0, “no problem,” to 4, “critical/meets diagnostic criteria”) of the severity of each of six different types of mental health problems: depression or sadness, traumatic stress, anxiety, suicidality, alcohol misuse, drug misuse, and behavior or conduct disorder. This rating correlates with Achenbach’s Teacher Report Form (24) at .68 and has a test-retest reliability of .86 A latent variable indicating the provider’s assessment of the youth’s environment was calculated from provider responses as to whether the youth had any of nine environmental problems: life stressors, family violence, problem peers, a violent neighborhood, violent school atmosphere, family instability, lack of family support, legal problems, or family financial problems. (The environmental aspects were not parallel to those asked of the youths.) A latent variable for providers’ resource knowledge was derived from ratings of 25 mental health resources. Knowledge of each resource was scored on a scale ranging from 0 to 4. Each of the four possible points per mental health resource represented a unit of knowledge in the following categories: familiarity with available services, referrals to resources, referPSYCHIATRIC SERVICES
Table 1
Cronbach’s alphas, means, and factor loadings for latent constructs to test whether the constructs will be used in structural equation modeling analysis and in the gateway provider modela Variable Youth’s report of mental health (Cronbach’s α=.77)b Suicidal symptoms Depression symptoms Posttraumatic stress disorder symptoms Conduct disorder symptoms Addiction symptoms Violent behaviors Youth’s perceived environment (Cronbach’s α=.68)b Drug dealing Shooting Murders Abandoned buildings Neighbors receiving welfare Homeless people in street Prostitution in neighborhood Provider’s resource knowledge (Cronbach’s α=.94)b Public health clinic School social worker or counselor Special schools Psychiatric hospital or inpatient unit Drug or alcohol inpatient Residential treatment Group or foster homes Emergency shelters Summer treatment or boarding Outpatient community mental health Psychologist, psychiatrist, counselor, or social worker Day treatment programs Family preservation services Emergency department for emotional problems Pediatrician or family doctor Self-help groups (for example, Alcoholics Anonymous) Crisis intervention services Social services Religious providers of services Traditional healers Talking circles Life skills Family or parenting programs Victims Other resources In-service training (Cronbach’s α=.94)b,c Alcoholism or drug abuse Assessment or diagnosis Intervention skills General mental health Adolescence Abuse or neglect Gangs or violence American-Indian issues Spiritual healing Any service (sum of referred or provided; Cronbach’s α=.78)b Individual or group counseling or therapy Substance abuse treatment Self-help groups or peer counseling Crisis intervention Inpatient or residential mental health Psychotropic medications Psychiatric evaluation Family or marital counseling Service coordination or casework
♦ ps.psychiatryonline.org ♦ August 2006 Vol. 57 No. 8
Mean score
Factor loading
.60 5.31 2.79 2.87 9.76 .82
.40 .45 .49 .71 .89 .61
.73 .49 .21 .23 .58 .40 .05
.63 .70 .55 .57 .30 .30 .30
.87 1.34 .56 .44 .49 .26 .62 .38 .22 .62 .93 .27 .18 .16 .40 .78 .54 .96 .78 .68 .38 .40 .76 .48 .38
.67 .55 .53 .71 .79 .66 .81 .63 .47 .79 .69 .63 .61 .55 .56 .74 .79 .71 .25 .26 .36 .54 .77 .77 .45 .84 .80 .76 .82 .84 .82 .86 .74 .64
.67 .08 .10 .09 .04 .08 .07 .07 .14
.45 .42 .39 .64 .59 .66 .77 .34 .59 Continues on next page
1187
Table 1
continued from previous page Variable Talked with youth Provider’s assessment of youth’s mental health (Cronbach’s α=.85)b Depression Traumatic stress Anxiety Suicidality Alcohol addiction Drug addiction Behavior or conduct disorder Presented withc Depression or sadness Suicidality Traumatic stress Behavior problems Alcohol issues Drug issues Provider’s perception of youth’s environment (Cronbach’s α=.76)b Life stressors Problem peers Violence in family Violent neighborhood Family instability Violent school Lack of family support Legal problems Financial problems a b c
Mean score
Factor loading
.74
.27
1.13 .56 .65 .29 .48 .46 .88
.74 .55 .60 .55 .73 .64 .22 .52 .42 .56 .41 .53 .45
.35 .36 .18 .22 .38 .22 .30 .15 .35
.56 .34 .54 .54 .62 .45 .63 .44 .46
Factor loadings represent how well the measured variable represents the underlying latent construct. Alpha is the variable’s consistency with the other items measuring the latent constructs. Items consisted of yes or no variables and thus did not yield meaningful means.
provider’s training were related to the provider’s assessment of the youth’s mental health and how the provider’s assessment of the youth and how the provider’s resource knowledge were related to services offered (26,27). The models were tested with weighted least-squares estimation in PROC CALIS (27) because of the ordinal nature of the variable indicating provider type (28). Paths with a beta of .10 or greater or with a significance level of .05 or less were retained in the model. The analyses followed a two-step procedure (20,26). In step one, latent variables were created from iterated principal factor analysis. After oblique rotation items loading on the first factor were retained, the single factor analysis was rerun to obtain factor loadings. Factors used as indicator variables reduce the random error and uniqueness associated with indicators measured as straight summary scores (28). Step two was the SEM analysis, which used the latent variables in PROC CALIS to test the fit between our model and the data. Each data point was constituted by a unique provider (N=188). Analyses of the data set without any repeating youths (N=141) resulted in the same significant support for the hypothesis.
Results rals from resources, and contacts with available services (19). On average, providers were familiar with a mean±SD of 11.8±7.1 resources. A latent variable for in-service training was operationalized from counts of topics covered in in-service training. The model was also tested for number of hours with no difference in significance. Thirty-four (36 percent) nonspecialist providers and 12 (87 percent) specialist providers received some type of in-service training in the past year. No traditional providers and only seven (10 percent) informal providers had any type of in-service training. Provider type was a manifest ordinal variable that was based on employment and training rather than the provider’s relation to the youth. A latent variable for services offered was developed from questions about directly offering, recommending, or 1188
referring to the following ten possible addictions or mental health services: individual or group counseling, substance abuse treatment, involvement in a self-help group, crisis intervention, inpatient or residential mental health treatment, medication, psychiatric evaluation, family or marital counseling, service coordination or referral, or a discussion with the youth about the problem. The informal providers offered a mean of 1.6±.6 different services; traditional providers, 1.0±.5 different services; nonspecialists, 2.2±2.3 different services; and specialists, 4.8±3.7 different services. All youths who used specialist providers were referred to them by informal or nonspecialist providers (5). Analyses We used SAS software for all analyses (25). SEM was used to test how the youth’s mental health and the PSYCHIATRIC SERVICES
As shown in Figure 2, SEM revealed that variance in services offered (recommended, referred, or directly provided) was predicted largely by gateway providers’ knowledge of youths’ mental health problems and of service resources. (Provider type was not significant in the SEM that used only nonrepeating youths.) In all, 30 percent of the variance in providers’ actions was predicted by providers’ assessment of youths’ addictions or mental health problems (.35), providers’ resource knowledge (.24), and provider type (.19). In turn, 38 percent of the variance in providers’ assessment of youths’ addictions or mental health problems was predicted by youths’ self-reported addictions or mental health problems (.33) and by providers’ assessment of youths’ environment (.46). Thirty-three percent of the variance in providers’ resource knowl-
♦ ps.psychiatryonline.org ♦ August 2006 Vol. 57 No. 8
Figure 2
Test of the gateway provider model by using data from 173 unique providers who served American-Indian youths and who were interviewed in the American Indian Multisector Help Inquiry Studya Youth constructs
Services
Provider constructs
Provider type
.17
.19
–.14
.43
Youth’s perceived environment
In-service training r2=.18
.53
.21
.42
Provider’s resource knowledge r2=.33
.16 Youth’s report of mental health r2=.28
.24 Services offered r2=.30
.28 .15
.33
Provider’s assessment of youth’s environment r2=.12
.35
.46 Provider’s assessment of youth’s mental health r2=.38 a
Adjusted goodness of fit index=.93; χ2=19.91, df=14, p
