Abstract
Objectives. We assessed the detection of mental illness in an adult population of substance abuse patients and the rate of referral for mental health treatment.
Methods. We obtained combined administrative records from 1994 to 1997 provided by the New Jersey substance abuse and mental health systems and estimated detection and referral rates of patients with co-occurring disorders (n = 47 379). Mental illness was considered detected if a diagnosis was in the record and considered undetected if a diagnosis was not in the record but the patient was seen in both treatment systems within the same 12-month period. Predictors of detection and referral were identified.
Results. The detection rate of co-occurring mental illness was 21.9% (n=10364); 57.9% (n=6001) of these individuals were referred for mental health treatment. Methadone maintenance clinics had the lowest detection rate but the highest referral rate. Male, Hispanic, and African American patients, as well as those who used heroin or were in the criminal justice system, had a higher risk of mental illness not being detected. Once detected, African American patients, heroin users, and patients in the criminal justice system were less likely to be referred for treatment.
Conclusions. There is a need to improve the detection of mental illness among substance abuse patients and to provide integrated treatment.
Individuals who have co-occurring mental illness and addiction disorders make up a significant, understudied population in mental health and substance abuse treatment systems. Several studies report a high prevalence of co-occurring disorders among mentally ill or substance abuse patients in the general US population.1–2 In 1996, the National Comorbidity Survey reported that 42.7% of individuals who had a 12-month addictive disorder had at least one 12-month mental health disorder.1 The more recent National Epidemiological Survey on Alcohol and Related Conditions found that 60.3% of respondents who had drug use disorders and who sought treatment had at least 1 independent mood disorder, and 42.6% had at least 1 independent anxiety disorder.3 Although substantial evidence shows that patients who have co-occurring mental illness and addiction disorders have more functional impairment, more behavioral problems, and a high risk of HIV or hepatitis infection,4–13 it is a challenge to detect the co-occurrence of these disorders.
Low rates of detecting co-occurring disorders in patients are supported by previous research. In a recent study, the New Jersey Division of Addiction Services (DAS) found the detection rates for patients who had co-occurring disorders to be 57% in the mental health treatment system and 23% in the addiction treatment system.14 Other studies report equally low rates. As cited by Drake et al.,15 Ananth et al. reported a 25% detection rate for patients in an acute care psychiatric setting.16 Among adolescents who receive mental health treatment, detection rates ranged from 57% in a continuum-of-care system, a mental health care network designed to be a highly integrated and well-coordinated community network that provides a full range of care to 4% in a traditional fee-for-service system.17
The underdiagnosis of co-occurring disorders in mental health and substance abuse populations has not received sufficient attention, given the importance of accurate diagnosis to effective treatment. In its report to Congress, the Substance Abuse and Mental Health Services Administration (SAMHSA) defined the elements of effective treatment for individuals who have co-occurring disorders: time-sensitive screening, comprehensive assessment, and program-oriented clinical interventions for medications and integrated psychosocial treatments.18 Recent studies have shown integrated treatment programs to be effective.19–21 However, when either substance abuse or mental health disorders are undiagnosed, patients lack access to integrated programs and tend to respond poorly to treatment interventions that emphasize only 1 behavioral disorder.
Although several studies have examined the factors that affect the detection of mental health problems in primary care or inpatient settings,22–25 few studies have systematically examined diagnostic accuracy across addiction treatment settings. Nor have many studies assessed both treatment system and patient factors that contribute to accurate diagnosis. This paper addresses the detection of co-occurring disorders within the New Jersey addiction treatment system. Specifically, the authors examine (1) the extent to which patients who have co-occurring mental illness and addiction disorders are diagnosed in an adult addiction treatment population, (2) the extent to which accurately diagnosed patients are referred for mental health treatment, and (3) patient and treatment characteristics associated with detection and referral for mental health treatment.
METHODS
Data
Our data are from the combined 1994–1997 administrative records of the Alcohol and Drug Abuse Data System (ADADS) and the Uniform Services Transaction Form (USTF), which was originally prepared by DAS for a study of treatment access and use of available services. These 2 data systems collect patient admission and discharge information from all publicly funded and most privately funded substance abuse and mental health treatment agencies in New Jersey. Although the data collection process differs among agencies, typically an intake worker or clinician collects information on the day of the patient’s admission and completes a discharge form within days of the patient’s discharge. The DAS maintains internal quality control procedures to ensure the accuracy and completeness of the records.
At the time these data were collected, New Jersey addiction agencies did not use a standardized mental health screening tool, and few agencies conducted systematic mental health screening. There was also substantial variability across agencies in levels of staff experience and training. For example, a recent study of New Jersey addiction treatment outcomes found that among the 20 participating inpatient, outpatient, and methadone agencies, there was a range of 0%–80% in the proportion of staff certified as addiction counselors.26
Both ADADS and USTF records contain patient demographic and treatment-related information. A detailed description of the data set, the process of compiling the data, and the identification of patients who had co-occurring disorders can be found in a previous report.14 Briefly, the DAS created the data set by randomly selecting 1 record per person (an index admission) from the total of each individual’s admissions to the addiction or mental health treatment systems between 1994 and 1997. Among index admissions to the addiction system, 130604 unique individuals were adults aged 18 years and older. A match was then sought for the index admission within the mental health data set in order to identify patients who had been treated in both systems. On the basis of this data-matching process, substance abuse patients were grouped into 2 categories: those with addiction problems only and those with co-occurring disorders. A patient was identified as having a co-occurring disorder if at least 1 of 2 criteria was met: (1) the patient had both an addiction and mental health diagnosis or a need for mental health treatment indicated in the index admission or discharge record or (2) the patient had an admission to the New Jersey mental health treatment system during the 12 months before or after the index addiction admission. By these criteria, 47379 adult substance abuse patients were identified as having a co-occurring mental illness. This was 36.3% of the adult patients who were treated in the New Jersey addiction treatment system between 1994 and 1997. Among the 47379 patients who had co-occurring disorders, only 8.6% met both criteria, and 13.3% met criterion (1) but not (2). Most of the patients who had co-occurring disorders (78.1%) met only criterion (2).
Dependent Variables
The addiction treatment provider was considered to have detected co-occurring mental illness only if criterion (1) was met. If only criterion (2) was met, the co-occurring disorder was considered to be undetected. The treatment referral was determined by whether or not a referral for mental health services was indicated on the discharge record.
Independent Variables
Because detection and referral are pathways to treatment for patients who have co-occurring disorders, our analysis included patient-centered variables that were associated with the use of mental health services or addiction treatment in several national studies. These variables include gender, age, race/ethnicity, education, family income, and reimbursement source.27–31 Age, race, and gender were associated with undetected mental illness among primary care patients,22 although homelessness and employment status were found to be predictive of a co-occurring disorder.32
We selected treatment characteristics for our analysis that were shown by previous research to have an impact on access to care. These included referral source, polydrug or monodrug use, and number of past drug treatment episodes.27,32 In addition, primary drug used, treatment setting at admission, and length (days) of treatment were included because of their significant association with detection and referral rates in preliminary bivariate analyses.
Treatment settings were grouped into five categories: (1) outpatient, including traditional or intensive outpatient care; (2) methadone, including outpatient methadone maintenance or detoxification; (3) short-term residential, typically 28 days; (4) long-term residential or halfway house, typically 180 days or longer; and (5) detoxification, including hospital or residential detoxification. The index admission year was included in the preliminary analysis to test time trends in detection and referral, but to promote a parsimonious model, it was dropped in the final analysis because it was statistically insignificant.
Data Analyses
We described the characteristics of patients who had co-occurring disorders and compared them to the characteristics of patients who had only substance use disorders. We then estimated the rates at which co-occurring disorders were detected by health care providers. Detection rates were compared among different patient and treatment characteristics, and the differences were tested using χ2 statistic and t tests. The associations between patient and treatment characteristics (independent variables) and detection (dependent variable) were further examined using a logistic regression model. Similar bivariate and logistic regression analyses were conducted for patients in whom mental illness had been detected to identify factors associated with referral to mental health treatment. We identified significant characteristics (at the .05 level) and reported the estimated odds ratios (ORs) and 95% confidence intervals. All statistical analyses were conducted using SAS version 8.0 (SAS Institute Inc.,Cary, NC).
RESULTS
Substance Abuse Patients With Co-Occurring Mental Illness
Patients who had co-occurring disorders were more likely to be female than patients who had only addiction disorders (33.2% vs 24.7%), but there was no difference between the groups with respect to age or race/ethnicity (Table 1 ▶). Patients who had co-occurring disorders were more likely than addiction-only patients to have Medicaid or Medicare coverage (18.0% vs10.4%), to be self-referred for treatment or to be referred by a friend or family member (50.2% vs 39.3%), and to use heroin (40% vs 28.2%). Patients who had co-occurring disorders were less likely than addiction-only patients to be treated in outpatient treatment settings (34.9% vs 52.3%) and to have had previous addiction treatment (31.1% vs 51.0%). The length of an index admission stay was shorter for patients who had co-occurring disorders than for addiction-only patients (88.3 vs 109.7 days). Index admissions for both groups were evenly distributed across the 4 years (1994–1997).
TABLE 1—
Demographic Characteristics of Patients Treated for Substance Abuse Problems: New Jersey Substance Abuse Treatment Systems, 1994–1997
| Co-Occurring Disorder, No. (%) | Substance Abuse Only, No. (%) | |
| All | 47 379 (100.0) | 83 225 (100.0) |
| Race/ethnicity | ||
| White | 24 547 (52.0) | 41 965 (50.6) |
| African American | 16 228 (34.4) | 28 773 (34.7) |
| Hispanic or Latino | 5 871 (12.4) | 10.961 (13.2) |
| Others | 577 (1.2) | 1 262 (1.5) |
| Gender | ||
| Female | 15 722 (33.2) | 20 531 (24.7) |
| Male | 31 657 (66.8) | 62 694 (75.3) |
| Age, ya | ||
| 18–34 | 27 057 (57.2) | 44 964 (54.0) |
| 35–44 | 14 790 (31.2) | 25 916 (31.1) |
| 45–54 | 4 258 (9.0) | 8 790 (10.6) |
| ≥55 | 1 274 (2.7) | 3 555 (4.3) |
| Homeless status | ||
| Yes | 383 (0.9) | 394 (0.5) |
| No | 44 074 (99.1) | 82 831 (99.5) |
| Employment status | ||
| Unemployed | 18 846 (39.9) | 27 820 (33.6) |
| Employed/not in force | 28 376 (60.1) | 55 081 (66.4) |
| Education | ||
| Less than high school | 16 142 (34.1) | 27 243 (32.7) |
| High school or GED | 21 993 (46.4) | 39 045 (46.9) |
| More than high school | 9 244 (19.5) | 16 937 (20.4) |
| Annual household incomeb | ||
| Poor | 41 768 (88.2) | 69 065 (83.0) |
| Low income | 2 768 (5.8) | 6 662 (8.0) |
| Middle income | 2 210 (4.7) | 5 767 (6.9) |
| High income | 633 (1.3) | 1 731 (2.1) |
| Reimbursement source | ||
| Medicaid/Medicare | 7 423 (18.0) | 7 724 (10.4) |
| Private | 6 574 (15.9) | 13 216 (17.8) |
| Self-pay | 7 818 (19.0) | 18 292 (24.7) |
| Nonec | 19 424 (47.1) | 34 910 (47.1) |
| Referred source | ||
| Self/family/friend | 23 694 (50.2) | 32 586 (39.3) |
| Other addiction providers | 5 689 (12.1) | 8 451 (10.2) |
| Mental health/medical providers | 8 488 (18.0) | 5 930 (7.2) |
| Legal | 5 896 (12.5) | 29 476 (35.6) |
| Others | 3 420 (7.3) | 6 443 (7.8) |
| Primary drug used | ||
| Alcohol | 16 874 (35.7) | 39 254 (47.3) |
| Heroin | 18 938 (40.1) | 23 414 (28.2) |
| Crack/cocaine | 7 885 (16.7) | 12 682 (15.3) |
| Marijuana | 2 200 (4.7) | 5 628 (6.8) |
| Others | 1 329 (2.8) | 1 934 (2.3) |
| Treatment settings at admissiond | ||
| Outpatient | 16 142 (34.9) | 42 793 (52.3) |
| Methadone | 8 596 (18.6) | 10 605 (13.0) |
| Short-term residential | 6 126 (13.2) | 5 879 (7.2) |
| Long-term residential | 2 598 (5.6) | 3 662 (4.5) |
| Detoxification | 12 855 (27.8) | 18 856 (23.1) |
| Polydrug use | ||
| Yes | 18 252 (59.8) | 42 565 (51.3) |
| No | 18 974 (40.2) | 40 347 (48.7) |
| Past addiction treatment episode | ||
| None | 14 747 (31.1) | 42 422 (51.0) |
| One | 12 375 (26.1) | 19 171 (23.0) |
| Two | 7 615 (16.1) | 8 523 (10.2) |
| Three | 4 240 (9.0) | 4 114 (4.9) |
| Four or more | 8 402 (17.7) | 8 995 (10.8) |
| Year | ||
| 1994 | 12 816 (27.1) | 21 477 (25.8) |
| 1995 | 11 564 (24.4) | 21 187 (25.5) |
| 1996 | 11 022 (23.3) | 20 910 (25.1) |
| 1997 | 11 977 (25.3) | 19 651 (23.6) |
| Length of treatment in days | ||
| Mean (SD) | 88.3 (178.3) | 109.7 (180.3) |
| Median | 27 | 52 |
Note. GED = general equivalency diploma.
aMean age of patients with co-occurring disorders = 33.8 years (9.2); mean age of patients with substance abuse only = 34.7 years (10.1).
bPoor < 125% federal poverty level (FPL); low income = 125%–199% FPL; middle income = 200%–399% FPL; high income ≥400% FPL.
cNo expected reimbursement source and did not self-pay for the treatment.
dOutpatient, including traditional outpatient or intensive outpatient care; methadone, including outpatient methadone maintenance and outpatient methadone detoxification; short-term residential care, typically 28 days; long-term residential care, typically 180 days or longer or halfway house; and detoxification, hospital detoxification, or residential detoxification.
Detection Rates
The overall rate of detection for co-occurring disorders was low. Co-occurring disorders were detected in only 21.9% of patients by their addiction treatment provider (Table 2 ▶). Providers detected disorders in more White patients (28.5%) than African Americans (13.9%) and Hispanics (16.2%) and in more women (26.0%) than men (19.8%). Disorders were more likely to be detected in older patients. By contrast, low rates of detection were found among patients who were unemployed, not homeless, less educated, or low income; who lacked a reimbursement source; or who had 1–3 previous treatment episodes. Most noticeably, patients who used heroin had the lowest detection rate (9.2%) compared with patients who used other substances (32.0% for alcohol, 23.9% for crack cocaine, and 33.8% for marijuana). Across all treatment settings, co-occurring disorders were detected in the lowest proportion of patients at methadone maintenance clinics (7.3%); disorders were detected in the highest proportion of patients at short-term residential programs (42.3%). The mean number of treatment days also was longer for patients who had disorders detected (mean = 119.2 days) compared with patients for whom disorders were not detected (mean = 78.2 days) (result not shown).
TABLE 2—
Odd Ratios (ORs) and 95% Confidence Intervals (CIs) for Substance Abuse Patients With Co-Occurring Mental Disorder Detected During Treatment: New Jersey Substance Abuse System, 1994–1997
| Logistic Regression: Co-Occurring Disorder Positively Detecteda | |||
| No. | Detected, %b | OR (95% CI) | |
| All | 47 379 | 21.9 | |
| Race/ethnicity | |||
| White | 24 547 | 28.5 | Referent |
| African American | 16 228 | 13.9 | 0.52* (0.48, 0.55) |
| Hispanic or Latino | 5 871 | 16.2 | 0.83* (0.76, 0.91) |
| Others | 577 | 24.1 | 0.97 (0.77, 0.24) |
| Gender | |||
| Female | 15 722 | 26.0 | 1.50* (1.41, 1.60) |
| Male | 31 657 | 19.8 | Referent |
| Age, y | |||
| Mean (SD) | 35.0 (10.3) | ||
| 18–34 | 27 057 | 20.1 | 0.90* (0.85, 0.95) |
| 35–44 | 14 790 | 22.4 | Referent |
| 45–54 | 4 258 | 27.7 | 1.19* (1.10, 1.28) |
| ≥55 | 1 274 | 34.5 | 0.93 (0.83, 1.04) |
| Homeless status | |||
| Homeless | 383 | 25.1 | 1.23 (0.89, 1.72) |
| Not homeless | 44 074 | 22.0 | Referent |
| Employment status | |||
| Unemployed | 18 846 | 15.8 | 0.67* (0.63, 0.72) |
| Employed/not-in-force | 28 376 | 25.9 | Referent |
| Education | |||
| Less than high school | 16 142 | 19.4 | 0.98 (0.94, 1.03) |
| High school or GED | 21 993 | 21.8 | 0.94* (0.90, 0.97) |
| More than high school | 9 244 | 26.4 | Referent |
| Annual household incomec | |||
| Poor | 41 768 | 21.8 | 1.08* (1.00, 1.16) |
| Low income | 2 768 | 21.5 | 0.97 (0.88, 1.08) |
| Middle income | 2 210 | 23.3 | 0.96 (0.87, 1.07) |
| High income | 633 | 26.1 | Referent |
| Reimbursement source | |||
| Private | 6 574 | 32.3 | Referent |
| Medicaid/Medicare | 7 423 | 26.8 | 1.45* (1.37, 1.54) |
| Self-pay | 7 818 | 24.8 | 1.11* (1.05, 1.18) |
| Noned | 19 424 | 17.6 | 0.70* (0.67, 0.73) |
| Referred source | |||
| Other addiction providers | 5 689 | 19.5 | Referent |
| Self/family/friend | 23 694 | 17.1 | 1.15* (1.09, 1.21) |
| Mental health/medical providers | 8 488 | 43.9 | 1.99* (1.86, 2.11) |
| Legal | 5 896 | 21.3 | 0.63* (0.59, 0.67) |
| Others | 3 420 | 22.3 | 0.90* (0.83, 0.99) |
| Primary drug used | |||
| Alcohol | 16 874 | 32.0 | Referent |
| Heroin | 18 938 | 9.2 | 0.47* (0.43, 0.50) |
| Crack/cocaine | 7 885 | 23.9 | 0.91* (0.85, 0.96) |
| Marijuana | 2 200 | 33.8 | 1.36* (1.24, 1.50) |
| Others | 1 329 | 41.8 | 1.54* (1.37, 1.73) |
| Treatment setting at admissione | |||
| Outpatient | 16 142 | 30.3 | Referent |
| Methadone | 8 596 | 7.3 | 0.35* (0.36, 0.39) |
| Short-term residential | 6 126 | 42.3 | 2.70* (2.53, 2.88) |
| Long-term residential | 2 598 | 16.0 | 1.14* (1.03, 1.27) |
| Detoxification | 12 855 | 11.2 | 0.58* (0.54, 0.62) |
| Polydrug use | |||
| Monodrug | 18 974 | 22.1 | Referent |
| Polydrug | 18 252 | 21.7 | 1.13* (1.06, 1.20) |
| Past addiction treatment episode | |||
| None | 14 747 | 25.2 | Referent |
| One | 12 375 | 18.3 | 0.82* (0.78, 0.87) |
| Two | 7 615 | 17.4 | 0.80* (0.753, 0.86) |
| Three | 4 240 | 17.7 | 0.92 (0.85, 1.00) |
| Four or more | 8 402 | 27.4 | 1.48* (1.40, 1.57) |
| Length of treatment in days | 1.00* (1.00, 1.00) | ||
| Mean (SD)f | 119.2 (219.4) | OR = 1.015 for every 10 days | |
| Median | 36.0 | ||
Note. GED = general equivalency diploma.
aLikelihood-ratio test statistic = 7671.04, df = 35, P < .0001.
bP values for all categories were <.001 except household income, where P = .020, and polydrug user status, where P = 0.244. P values were based on χ2 tests for the association between percentage identified and patient characteristics.
cPoor < 125% federal poverty level (FPL); low income = 125%–199% FPL; middle income = 200%–399% FPL; high income ≥400% FPL.
dNo expected reimbursement source and did not self-pay for the treatment.
eOutpatient, including traditional outpatient or intensive outpatient care; methadone, including outpatient methadone maintenance and outpatient methadone detoxification; short-term residential care, typically 28 days; long-term residential care, typically 180 days or longer or halfway house; and detoxification, hospital detoxification, or residential detoxification.
fSignificantly more days of treatment for patients whose mental illness was detected than those not detected, P < .001.
*Statistically significant at α = 0.05.
Characteristics Associated With Detection
Co-occurring disorders were less likely to be detected in African American (OR = 0.52) and Hispanic individuals (OR = 0.83) than in Whites (Table 2 ▶). Disorders in women were 1.503 times as likely to be detected as in men. Co-occurring disorders were about 10% less likely to be detected in patients younger than 35 years and were 19% more likely to be detected in older patients (45 to 54 years), than in patients between 35 and 44 years.
The probability of detection also was influenced by socioeconomic factors. Disorders were 0.67 times as likely to be detected in unemployed patients as in patients who were employed. Disorders in high-school graduates were less likely to be detected (OR = 0.94) than in patients who had some college education or better. Patients in the poor income group were approximately 8% more likely to have disorders detected than were high-income patients. Compared with privately insured patients, Medicaid and Medicare (OR = 1.45) and self-pay (OR = 1.11) patients were more likely to have disorders detected, but those who had no source of reimbursement were only 0.70 times as likely to have disorders detected.
Compared with primary alcohol users, primary heroin or crack and cocaine users were significantly less likely to have disorders detected (OR = 0.47 for heroin; OR = 0.91 for crack and cocaine); primary marijuana or other drug users were more likely to have disorders detected. Disorders in polydrug users were 1.13 times more likely to be detected than in monodrug users.
With respect to treatment-related characteristics, patients who were referred from the criminal justice system were significantly less likely to have disorders detected (OR = 0.63), and those patients referred from medical or mental health settings were significantly more likely to have disorders detected (OR = 1.99) than patients who were referred for treatment from other addiction agencies. By treatment setting, patients in methadone maintenance clinics or hospital/residential detoxification programs had the lowest (OR = 0.35 and OR = 0.58, respectively), and patients in short-term residential programs had the highest (OR = 2.70), odds of disorders being detected compared with outpatients. The odds that disorders would be detected were higher for patients who had 4 or more previous addiction treatment episodes (OR = 1.48) than for first-time admissions; however, patients who had 1–3 previous episodes were significantly less likely to have disorders detected. Also, the odds of detection increased by 2% for every 10 days the patient spent in treatment.
Referral Rates
Among the 10364 patients in whom co-occurring disorders were detected, 57.9% were referred for mental health treatment (Table 3 ▶). Referral rates differed by age, income, reimbursement source, referral source, primary drug used, and treatment setting at admission. Also, patients who were referred had shorter addiction treatment stays than patients who were not referred (mean = 105.3 vs 135.7 days) (result not shown).
TABLE 3—
Odd Ratios (ORs) and 95% Confidence Intervals (CIs) for Substance Abuse Patients With Co-Occurring Mental Disorder Detected During Treatment: New Jersey Substance Abuse System, 1994–1997
| No. | Referred to Mental Health Treatment, % | Pa | Logistic Regression: Referred to Mental Health Treatment, OR (95% CI)b | |
| All | 10 364 | 57.9 | ||
| Race/ethnicity | .129 | |||
| White | 6 991 | 58.5 | Referent | |
| African American | 2 249 | 55.8 | 0.85* (0.76, 0.96) | |
| Hispanic or Latino | 949 | 58.9 | 1.09 (0.93, 1.28) | |
| Others | 139 | 55.4 | 0.98 (0.67, 1.43) | |
| Gender | .420 | |||
| Female | 4 090 | 58.4 | 1.03 (0.94, 1.13) | |
| Male | 6 274 | 57.6 | Referent | |
| Age, y | <.001 | |||
| Mean (SD) | 35.3 (10.6) | |||
| 18–34 | 5 429 | 56.1 | 0.87* (0.80, 0.95) | |
| 35–44 | 3 316 | 59.2 | Referent | |
| 45–54 | 1 180 | 59.8 | 1.02 (0.91, 1.14) | |
| ≥55 | 439 | 65.2 | 1.17 (0.98, 1.39) | |
| Homeless status | .616 | |||
| Homeless | 96 | 60.4 | 0.86 (0.50, 1.47) | |
| Not homeless | 9 674 | 57.7 | Referent | |
| Employment status | .887 | |||
| Unemployed | 2 976 | 58.0 | 0.99 (0.89, 1.09) | |
| Employed/not-in-force | 7 349 | 57.8 | Referent | |
| Education | .092 | |||
| Less than high school | 3 137 | 56.3 | 0.95 (0.88, 1.01) | |
| High school or GED | 4 783 | 58.6 | 1.03 (0.97, 1.10) | |
| Some college or more | 2 444 | 58.6 | Referent | |
| Annual household incomec | <.001 | |||
| Poor | 9 088 | 59.7 | 1.36* (1.21, 1.52) | |
| Low income | 595 | 46.1 | 0.96 (0.82, 1.13) | |
| Middle income | 516 | 44.4 | 0.85 (0.72, 1.01) | |
| High income | 165 | 46.7 | Referent | |
| Reimbursement source | <.001 | |||
| Private | 2 121 | 65.5 | Referent | |
| Medicaid/Medicare | 1 990 | 62.0 | 0.99 (0.91, 1.08) | |
| Self-pay | 1 936 | 49.6 | 0.87* (0.80, 0.95) | |
| Noned | 3 409 | 56.8 | 0.92* (0.85, 0.99) | |
| Referred source | <.001 | |||
| Other addiction providers | 1 110 | 53.5 | Referent | |
| Self/family/friend | 4 046 | 59.3 | 0.93 (0.86, 1.01) | |
| Mental health/medical providers | 2 591 | 66.3 | 1.28* (1.17, 1.41) | |
| Legal | 1 804 | 46.5 | 0.89* (0.80, 0.98) | |
| Others | 763 | 54.7 | 1.01 (0.88, 1.16) | |
| Primary drug used | .001 | |||
| Alcohol | 5 401 | 57.6 | Referent | |
| Heroin | 1 734 | 61.1 | 0.83* (0.73, 0.94) | |
| Cocaine/crack | 1 887 | 55.6 | 0.94 (0.85, 1.05) | |
| Marijuana | 743 | 55.3 | 1.09 (0.94, 1.25) | |
| Others | 556 | 63.3 | 1.18 (1.00, 1.39) | |
| Treatment setting at admissione | <.001 | |||
| Outpatient | 2 310 | 47.3 | Referent | |
| Methadone | 624 | 75.0 | 2.60* (2.13, 3.18) | |
| Short-term residential | 2 591 | 68.3 | 1.13* (1.01, 1.25) | |
| Long-term residential | 415 | 50.0 | 0.70* (0.58, 0.84) | |
| Detoxification | 1 445 | 62.7 | 0.91 (0.80, 1.02) | |
| Polydrug use | .090 | |||
| Monodrug | 4 198 | 57.0 | Referent | |
| Polydrug | 6 123 | 58.6 | 1.12* (1.01, 1.23) | |
| Past addiction treatment episode | .212 | |||
| None | 3 723 | 56.4 | Referent | |
| One | 2 264 | 58.8 | 1.08 (0.98, 1.18) | |
| Two | 1 329 | 58.8 | 1.01 (0.909, 1.128) | |
| Three | 749 | 59.6 | 0.97 (0.844, 1.112) | |
| Four or more | 2 299 | 58.4 | 1.00 (0.91, 1.09) | |
| Length of treatment in days | 1.00* (1.00, 1.00) | |||
| Mean (SD)f | 105.3 (228.6) | OR = 0.966 for every 10 days | ||
| Median | 28.0 |
Note. GED = general equivalency diploma.
aBased on χ2 tests for the association between percentage who were referred and the patient’s characteristics.
bLikelihood-ratio test statistic = 643.18, df = 35, P < .0001.
cPoor < 125% federal poverty level (FPL); low income = 125%–199% FPL; middle income = 200%–399% FPL; high income ≥400% FPL.
dNo expected reimbursement source and did not self-pay for the treatment.
eOutpatient, including traditional outpatient or intensive outpatient care; methadone, including outpatient methadone maintenance and outpatient methadone detoxification; short-term residential care, typically 28 days; long-term residential care, typically 180 days or longer, or halfway house; and detoxification, hospital detoxification, or residential detoxification.
fSignificantly fewer days of treatment for patients with identified co-occurring disorder who were referred than those not referred, P < .001.
*Statistically significant at α = 0.05.
Characteristics Associated With Referral Rates
Overall, treatment and insurance-related characteristics appear to have played a more significant role in the referral for mental health treatment than other patient characteristics (Table 3 ▶). Patients who were referred for addiction treatment from the legal system had lower odds of being referred (OR = 0.89) for mental health treatment than those referred from other addiction providers. By treatment setting, methadone maintenance clinics were substantially more likely (OR = 2.60) to refer patients for mental health treatment than were outpatient programs. By contrast, long-term residential programs had a low referral rate (OR = 0.70). Unlike the positive effect of longer treatment programs on detection of disorders, longer stays were negatively associated with referral for mental health treatment. With respect to patient characteristics, African American patients were less likely to be referred for treatment (OR = 0.85) than White patients. The odds of referral for younger patients (age between 18 and 34) were 0.87 times those of patients between 35 and 44. Compared with high-income patients, individuals in the poor income group had higher odds of being referred for mental health treatment (OR = 1.36). Self-pay patients or those with no reimbursement source were less likely to be referred for treatment (OR = 0.87 and OR = 0.92, respectively). In terms of drug use, patients who used heroin had lower odds of being referred for mental health treatment (OR = 0.83) than alcohol users; polydrug users were 12% more likely to be referred.
DISCUSSION
Our findings are reflective of practices in a single state and, as such, are not nationally representative. In addition, these data reflect treatment practices that were in place in New Jersey until 2003, before system-wide implementation of a standardized mental health screening tool. Therefore, the findings are only suggestive of practices in states that do not have standardized mental health screening.
As with many studies that use administrative data, this study is subject to the accuracy and completeness of the records. Because DAS had a quality control system that systematically checked for and corrected missing data, it was expected that errors in the data set would be random and, given the size of the data set, that overall trends would be reliable. Because the identification of patients who had co-occurring disorders is based solely on treatment records, the findings are likely to underrepresent the true prevalence of co-occurring disorders. Many patients may not have sought mental health treatment or accessed treatment from nonreporting health care providers (such as the Veterans Administration, family doctors, or out-of-state providers). Finally, the method by which the original data set was compiled placed primary emphasis on treatment and patient characteristics identified in the patient’s index admission setting. Because all patients in this study had their index admission in the addiction system, the data exclude key mental health system variables, such as psychiatric diagnosis, that would have been desirable to include in the analysis.
Our findings revealed low detection rates for disorders and low referral rates for addiction treatment in patients who had co-occurring mental illness. The study also identified specific treatment settings and patient populations that should be targeted for systemic improvements. These findings are indicative of a larger national problem, which has prompted researchers and clinicians to identify the assessment and diagnosis of co-occurring disorders as an important area for new funding to provide training and services.33
A number of sociodemographic characteristics were significantly associated with detection of disorders and referral for treatment. Males, young adults,18–34 minorities (African Americans and Hispanics), and unemployed patients were at higher risk than other patients that their disorders would go undetected. African American patients, in particular, were about half as likely as White patients to be diagnosed and were less likely to be referred for treatment.
These differences among gender, age, and race/ethnicity are consistent with findings of studies that examined detection and treatment access in primary care and other populations.22,29–31 The greater risk of nondetection for minorities may relate, in part, to cultural differences in the presentation of symptoms,34,35 and lower detection rates may partially explain lower minority treatment access. In our study, however, even when disorders were diagnosed in African Americans, they were less likely than Whites to be referred for treatment.
Reimbursement source was also an important factor in this study. Patients who had no reimbursement source were at a significant disadvantage compared with those who had private insurance, Medicaid, or Medicare. Medicaid and Medicare patients were somewhat more likely than privately insured patients to be identified as having co-occurring disorders, possibly because they presented with more severe symptoms at admission, which made them more readily identifiable. Because public funding covers mental health treatment, however, we would expect the referral rates to be comparable.
The importance of drug-use characteristics was evidenced by the lower detection and referral rates for patients who were primary heroin and polydrug users. Patients who are polydrug users often present with more complex withdrawal symptoms than patients who are monodrug users. This increases the clinical challenge of differentiating symptoms of mental illness from symptoms of drug use. Similarly, patients who use heroin have high rates of antisocial personality disorder,36,37 which could cause clinicians to misinterpret complaints about mental health symptoms as manipulative behavior.
Across all treatment settings, methadone maintenance clinics had the lowest rate of detecting disorders but the highest rate of referrals for treatment. However, co-morbidity rates for patients who were receiving methadone maintenance treatment (46%) were similar to patients being treated in long-term residential (42%) and detoxification settings (41%). Because many New Jersey methadone clinics have large patient populations, excessive staff workloads might lead to low detection rates where only the sickest patients are identified. As a result, referral rates might be expected to be higher if the identified populations are more severely impaired. Detoxification settings also had low rates of identifying disorders. This was a predictable finding because such settings manage patients in acute withdrawal for limited time periods, which makes it difficult to differentiate mental health symptoms from withdrawal symptoms. By contrast, short-term residential programs were approximately 2.7 times as likely as outpatient programs to detect co-occurring disorders. This possibly reflects the fact that short-term residential programs adhere to a medical treatment model and have highly professional staff.26
As might be expected, disorders were identified more often in patients who were referred from mental health and medical sources, probably because those patients entered into addiction treatment with a preexisting mental health diagnosis. Similarly, self-or family-referred patients also experienced higher rates of diagnosis, which suggests that those patients may enter treatment with greater acknowledgment of their problems than patients who enter through other pathways. Patients who were referred for treatment by the legal system had significantly low detection rates, an alarming finding in light of US Department of Justice estimates that 72% of mentally ill prisoners have an addiction disorder.13,38 New Jersey has both drug and mental health courts, which divert individuals who have behavioral problems to treatment; however, these 2 court systems have separate discharge and re-entry planning processes, which could result in fragmented care for people who have dual problems. Also, many individuals who are referred for treatment by drug court have antisocial and other personality disorders,37,39 which can lead to the same risk experienced by heroin users; reports of depression and anxiety may be dismissed as unfounded.
Because of the large sample size of this study, many variables were found to be statistically significant even though the coefficients were small. However, several relationships that showed substantial effects have important implications for public policy. The low rates of disorder detection for African Americans, for example, point to racial disparities in health care access that require a greater public investment in creative and effective solutions. Similarly, the greater detection of mental illness in patients covered by Medicaid rather than privately insured patients points to a potentially greater public, as opposed to private, funding burden for treating co-occurring disorders, at least among patients in public addiction treatment programs. Finally, the low probability of detecting mental illness in people who use heroin, affects both heroin users and methadone clinics, the most common treatment setting, and suggests the need for further research into the factors unique to heroin use and the clinical processes of methadone clinics that might discourage accurate diagnoses.
Although structured psychiatric interviews significantly improve psychiatric diagnosis,40 at the time this data set was collected, standardized instruments were not commonly used by states to evaluate a patient’s mental status. However, New Jersey subsequently instituted a modified Addiction Severity Index (ASI)41 as a standardized intake instrument for all publicly funded treatment agencies. The ASI will help alert clinicians to possible mental health problems; however, other instruments, such as the Beck Depression Inventory-II42 or Mental Health Screening Form-III,43 would be useful enhancements for patients who present with possible psychiatric involvement. Improving the detection of co-occurring disorders, however, should be one component of a broader national goal to promote truly integrated treatment through large-scale systems change. As SAMHSA has indicated, this will require (1) cross-training of addiction and mental health professionals in screening, assessment, and specialized integrated treatment techniques; (2) complementary licensing and certification for addiction and mental health programs and staff; and (3) more flexible reimbursement mechanisms for financing the treatment of co-occurring disorders.
Acknowledgments
This work was partially supported by the Mental Health Services and Systems Training Program (grant T32 MH016242).
The authors thank Allan Horwitz and David Mechanic of the Institute for Health, Health Care Policy and Aging Research at Rutgers, the State University of New Jersey, for their comments.
Note. The findings and opinions reported here are those of the authors and do not necessarily represent the views of any other individuals or organizations.
Human Participant Protection This paper is one component of a study (Diagnosis and Treatment of Patients with Comorbid Mental Illness and Substance Use in the State of New Jersey) that has been approved by the institutional review board of Rutgers, The State University of New Jersey, as exempt.
Peer Reviewed
Contributors All authors helped to conceptualize ideas, interpret findings, review drafts of the article, and approve the final version. Hsou Mei Hu synthesized the analyses and led the writing. Anna Kline supervised the design and analysis of the study and interpretation of the findings. Fredrick Huang assisted with the analyses and interpretation of the findings. Douglas Ziedonis provided clinical expertise and supervised all aspects of the study.
References
- 1.Kessler RC, Nelson CB, McGonagle KA, Edlund MJ, Frank RG, Leaf PJ. The epidemiology of co-occurring addictive and mental disorders: implications for prevention and service utilization. Am J Orthopsychiatry. 1996;66:17–31. [DOI] [PubMed] [Google Scholar]
- 2.Results from the 2001 National Household Survey on Drug Abuse: Volume I. Summary of National Findings. Rockville, Md: Substance Abuse and Mental Health Services Administration, Office of Applied Studies; 2002. DHHS Publication SMA 02–3758.
- 3.Grant BF, Stinson FS, Dawson DA, et al. Prevalence and co-occurrence of substance use disorders and independent mood and anxiety disorders: results from the National Epidemiologic Survey on Alcohol and Related Conditions. Arch Gen Psychiatry. 2004; 61:807–816. [DOI] [PubMed] [Google Scholar]
- 4.King RD, Gaines LS, Lambert EW, Summerfelt WT, Bickman L. The co-occurrence of psychiatric and substance use diagnoses in adolescents in different service systems: frequency, recognition, cost, and outcomes. J Behav Health Serv Res. 2000;27:417–430. [DOI] [PubMed] [Google Scholar]
- 5.Steadman HJ, Mulvey EP, Monahan J, et al. Violence by people discharged from acute psychiatric inpatient facilities and by others in the same neighborhoods. Arch Gen Psychiatry. 1998;55:393–401. [DOI] [PubMed] [Google Scholar]
- 6.Abram KM, Teplin LA. Co-occurring disorders among mentally ill jail detainees. Implications for public policy. Am Psychol. 1991;46:1036–1045. [DOI] [PubMed] [Google Scholar]
- 7.Abram KM, Teplin LA, McClelland GM. Comorbidity of severe psychiatric disorders and substance use disorders among women in jail. Am J Psychiatry. 2003; 160:1007–1010. [DOI] [PubMed] [Google Scholar]
- 8.Abram KM, Teplin LA, McClelland GM, Dulcan MK. Comorbid psychiatric disorders in youth in juvenile detention. Arch Gen Psychiatry. 2003;60:1097–1108. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Coodin S, Staley D, Cortens B, Desrochers R, McLandress S. Patient factors associated with missed appointments in persons with schizophrenia. Can J Psychiatry. 2004;49:145–148. [DOI] [PubMed] [Google Scholar]
- 10.Hudson TJ, Owen RR, Thrush CR, et al. A pilot study of barriers to medication adherence in schizophrenia. J Clin Psychiatry. 2004;65:211–216. [DOI] [PubMed] [Google Scholar]
- 11.Rosenberg SD. Prevalence of HIV, hepatitis B, and hepatitis C in people with severe mental illness. Am J Public Health. 2001;91:31–37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Swartz MS, Swanson JW, Hiday VA, Borum R, Wagner HR, Burns BJ. Violence and severe mental illness: the effects of substance abuse and nonadherence to medication. Am J Psychiatry. 1998;155:226–231. [DOI] [PubMed] [Google Scholar]
- 13.The role of co-occurring substance abuse and mental illness in violence: workshop summary. Washington, DC: Institute of Medicine, Division of Neuroscience and Behavioral Health; 1999.
- 14.Kline A, Mammo A., Culleton R., Crowley C., Barton C. National Evaluation Data: Treatment Access and Service Utilization by Clients in the Substance Abuse and Mental Health Treatment Systems System. Trenton, NJ: New Jersey Department of Health and Senior Services, Division of Addiction Services; 2002.
- 15.Drake RE, McLaughlin P, Pepper B, Minkoff K. Dual diagnosis of major mental illness and substance disorder: an overview. New Dir Ment Health Serv. 1991;50:3–12. [DOI] [PubMed] [Google Scholar]
- 16.Ananth J, Vandewater S, Kamal M, Brodsky A, Gamal R, Miller M. Missed diagnosis of substance abuse in psychiatric patients. Hosp Community Psychiatry. 1989;40:297–299. [DOI] [PubMed] [Google Scholar]
- 17.King RD, Gaines LS, Lambert EW, Summerfelt WT, Bickman L. The co-occurrence of psychiatric and substance use diagnoses in adolescents in different service systems: frequency, recognition, cost, and outcomes. J Behav Health Serv Res. 2000;27:417–430. [DOI] [PubMed] [Google Scholar]
- 18.Report to Congress on the Prevention and Treatment of Co-Occurring Substance Abuse Disorders and Mental Disorders. Rockville, Md: US Department of Health and Human Services, Substance Abuse and Mental Health Services Administration. Available at: http://alt.samhsa.gov/reports/congress2002/index.html. Accessed December 2004.
- 19.Durell J, Lechtenberg B, Corse S, Frances RJ. Intensive case management of persons with chronic mental illness who abuse substances. Hosp Community Psychiatry. 1993;44:415–416, 428. [DOI] [PubMed] [Google Scholar]
- 20.Drake RE, Yovetich NA, Bebout RR, Harris M, McHugo GJ. Integrated treatment for dually diagnosed homeless adults. J Nerv Ment Dis. 1997;185:298–305. [DOI] [PubMed] [Google Scholar]
- 21.Drake RE, Essock SM, Shaner A, et al. Implementing dual diagnosis services for clients with severe mental illness. Psychiatr Serv. 2001;52:469–476. [DOI] [PubMed] [Google Scholar]
- 22.Borowsky SJ, Rubenstein LV, Meredith LS, Camp P, Jackson-Trihe M, Wells KB. Who is at risk of nondetection of mental health problems in primary care? J Gen Intern Med. 2000;15:381–388. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Smith MV, Rosenheck RA, Cavaleri MA, Howell HB, Poschman K, Yonkers KA. Screening for and detection of depression, panic disorder, and PTSD in public-sector obstetric clinics. Psychiatr Serv. 2004; 55:407–414. [DOI] [PubMed] [Google Scholar]
- 24.Fink P, Orbol E, Hansen MS, Sondergaard L, De Jonge P. Detecting mental disorders in general hospitals by the SCL-8 scale. J Psychosom Res. 2004;56:371–375. [DOI] [PubMed] [Google Scholar]
- 25.Zatzick D, Russo J, Rivara F, Roy-Byrne P, Jurkovich G, Katon W. The detection and treatment of posttraumatic distress and substance intoxication in the acute care inpatient setting. Gen Hosp Psychiatry. 2005; 27:57–62. [DOI] [PubMed] [Google Scholar]
- 26.Kline A, Barton C, Bruzios C, Merrill J, Kyllingstad A, Violette N. The New Jersey Treatment Outcome and Performance Pilot Study II: Final Project Report. Trenton, NJ: New Jersey Department of Health and Senior Services; 2003.
- 27.Green-Hennessy S. Factors associated with receipt of behavioral health services among persons with substance dependence. Psychiatr Serv. 2002;53:1592–1598. [DOI] [PubMed] [Google Scholar]
- 28.Watkins KE, Burnam A, Kung F, Paddock S. A national survey of care for persons with co-occurring mental and substance use disorders. Psychiatr Serv. 2001;52:1062–1068. [DOI] [PubMed] [Google Scholar]
- 29.Wells K, Klap R, Koike A, Sherbourne C. Ethnic disparities in unmet need for alcoholism, drug abuse, and mental health care. Am J Psychiatry. 2001;158: 2027–2032. [DOI] [PubMed] [Google Scholar]
- 30.Wang PS, Demler O, Kessler RC. Adequacy of treatment for serious mental illness in the United States. Am J Public Health. 2002;92:92–98. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Wang PS, Lane M, Olfson M, Pincus HA, Wells KB, Kessler RC. Twelve-month use of mental health services in the United States: results from the National Comorbidity Survey Replication. Arch Gen Psychiatry. 2005;62:629–640. [DOI] [PubMed] [Google Scholar]
- 32.Johnson ME, Brems C, Burke S. Recognizing co-morbidity among drug users in treatment. Am J Drug Alcohol Abuse. 2002;28:243–261. [DOI] [PubMed] [Google Scholar]
- 33.Ziedonis DM. Integrated treatment of co-occurring mental illness and addiction: clinical intervention, program, and system perspectives. CNS Spectr. 2004;9: 892–904, 925. [DOI] [PubMed] [Google Scholar]
- 34.Neighbors HW, Jackson JS, Campbell L, Williams D. The influence of racial factors on psychiatric diagnosis: a review and suggestions for research. Community Ment Health J. 1989;25:301–311. [DOI] [PubMed] [Google Scholar]
- 35.Baker FM, Bell CC. Issues in the psychiatric treatment of African Americans. Psychiatr Serv. 1999; 50:362–368. [DOI] [PubMed] [Google Scholar]
- 36.King VL, Kidorf MS, Stoller KB, Carter JA, Brooner RK. Influence of antisocial personality subtypes on drug abuse treatment response. J Nerv Ment Dis. 2001;189:593–601. [DOI] [PubMed] [Google Scholar]
- 37.Chen CC, Tsai SY, Su LW, Yang TW, Tsai CJ, Hwu HG. Psychiatric co-morbidity among male heroin addicts: differences between hospital and incarcerated subjects in Taiwan. Addiction. 1999;94:825–832. [DOI] [PubMed] [Google Scholar]
- 38.Ditton PM. Mental Health and Treatment of Inmates and Probationers. Washington, DC: US Department of Justice; 1999.
- 39.Swartz JA, Lurigio AJ. Psychiatric illness and co-morbidity among adult male jail detainees in drug treatment. Psychiatr Serv. 1999;50:1628–1630. [DOI] [PubMed] [Google Scholar]
- 40.Kranzler HR, Kadden RM, Burleson JA, Babor TF, Apter A, Rounsaville BJ. Validity of psychiatric diagnoses in patients with substance use disorders: is the interview more important than the interviewer? Compr Psychiatry. 1995;36:278–288. [DOI] [PubMed] [Google Scholar]
- 41.McLellan AT, Luborsky L, O’Brien CP, Woody GE. An improved diagnostic instrument for substance abuse patients: The Addiction Severity Index. J Nerv Ment Dis. 1980;168:26–33. [DOI] [PubMed] [Google Scholar]
- 42.Beck AT, Steer RA, Brown GK. BDI-II, Beck Depression Inventory: Manual. 2nd ed. San Antonio, Tex: Psychological Corp.; 1996.
- 43.Substance Abuse Treatment for Persons with Co-Occurring Disorders. Treatment Improvement Protocol (TIP) Series 42. Rockville, Md: Substance Abuse and Mental Health Services Administration, Center for Substance Abuse Treatment; 2005. DHHS Publication SMA 05–3992. [PubMed]