Transcript Document

The ability of Global Appraisal of Individual Needs (GAIN) Scales in Predicting Clinical Psychiatric Diagnoses
among Adolescents in Substance Abuse Treatment
Geetha Subramaniam1, Patricia Shane2, Michael L. Dennis3, Melissa Ives3, Lora Passetti3
1Johns
Hopkins University School of Medicine, Baltimore, MD; 2 Alta Bates Summit Medical Center, Sutter Health, Oakland, CA; 3Chestnut Health Systems, Bloomington, IL
Introduction
Results
Research consistently suggests over 70% of adolescents seeking treatment for a substance use disorder
(SUD) present with a co-occurring psychiatric disorder. Yet only 28% of public treatment records identified a
co-occurring mental health disorder. This poster explores the validity and contribution of a widely used
comprehensive intake assessment in improving identification of co-occurring psychiatric disorders among
adolescents entering SUD treatment.
10%
20%
30%
40%
50%
Any Internal or External Disorder 6% 9%
Any Internalizing Disorder
Methods
Clinical Diagnoses: All clinical psychiatric
diagnoses were made based on DSM-IV by a
qualified clinician (e.g. psychiatrist,
psychologist) as per usual standard for care at
each of the sites. These were made
independently of the GAIN (described below).
Figure 1. Identification of Disorders by Source (s)
0%
18%
Major Depressive Disorder
60%
70%
Table 2. Comparison of Clinical Records & GAIN
80%
90% 100%
96%
81%
12%
21%
54%
14%
84%
42%
76%
Table 1. Sample Characteristics Overall and by Site
Total
Characteristic
(n=697)
Age - Mean (S.D.)
15.9 (1.2)
69%
Male - %
Race: African American - %
25%
GAIN: The Global Appraisal of Individual
Needs, a comprehensive biopsychosocial
Caucasian - %
69%
structured interview, was administered at
Single Parent Family - %
47%
treatment intake by interviewers certified in it
In Juvenile/Criminal Justice System
68%
use. The GAIN incorporates mental health
12.2 (2.2)
Age of first use - Mean (S.D.)
diagnostic scales related to internalizing
83%
Weekly or more frequent substance use - %
disorders (e.g. depression, anxiety, trauma) and
84%
Lifetime
Dependence
-%
externalizing disorders (ADHD, conduct
53%
problems) that can be used as symptom counts Any Prior Substance Abuse Treatment - %
84%
Entering Residential Treatment - %
or to approximate a diagnosis.
75%
Lifetime Victimization
Sites and Participants: Participants were
60%
Prior mental health treatment history - %
recruited between 1999 and 2006 as part of 5
31%
research projects conducted at three sites: 106 Internalizing: Internal Mental Distress Scale (IMDS)\a
46%
Externalizing: Behavior Complexity Scale (BCS)\b
from Mountain Manor Treatment Center
38%
Both Internalizing and Externalizing (IMDS +BCS)
(MMTC), Baltimore, MD: 198 from Chestnut
Health Systems (CHS), Bloomington, IL; and
\a Percent of 41 past year symptoms related to somatic,
393 from Thunder Road/ Sutter Health,
depression, anxiety, and trauma disorders
Oakland, CA. Table 1 shows the characteristics \b Percent of 33 past year symptoms related to inattentiveness,
of study participants. The bottom three rows are
hyperactivity, impulsivity, and conduct disorders
mental health symptom counts based on self
reports using the GAIN.
Statistical Analyses: The sensitivity (i.e., percent of cases correctly identified) of each method was evaluated
relative to the combined approach. The ability of the GAIN to use self reported data to identify who is likely to
get a diagnosis if they saw a clinician was tested using a simple odds ratio.
PTSD/Acute Stress Disorder 3%
29%
Other Anxiety Disorder 5%
11% 43%
27%
Any Externalizing Disorder 5%
31%
ADHD Related 8%
57%
43%
Conduct Disorder Related 6%
21%
37%
Both Internal & External Dis.
0%
4% 37%
37%
10%
20%
Clinical Records Only
73%
42%
14%
30%
93%
40%
85%
10% 61%
50%
60%
GAIN Only
70%
Sensitivity \a
Clinical
Records GAIN
90%
94%
Any Internal or External Disorder
86%
78%
Any Internalizing Disorder
82%
72%
Major Depressive Disorder
32%
93%
PTSD/Acute Stress Disorder
26%
86%
Other Anxiety Disorder
67%
95%
Any Externalizing Disorder
ADHD Related
Conduct Disorder Related
Both Internal and External Disorder
40%
56%
77%
89%
93%
44%
Odds Ratio\b
OR (95% CI)
4.9
(4.4 - 5.5)
3.9
3.5
6.4
1.9
(3.5 - 4.2)
(3.1 - 3.8)
(5.9 - 6.9)
(1.4 - 2.4)
2.7
1.6
2.8
0.8
(2.3 - 3.2)
(1.3 – 2.0)
(2.4 - 3.2)
(0.4 – 1.3)
\a Sensitivity is the percent of "diagnoses" identified by each method
relative to those identified by either method.
\b Odds Ratio of the extent to which the GAIN predicts increased likelihood
of a clinical diagnosis and 95% confidence interval (CI)
80%
90% 100%
Both
Figure 1 shows that having any co-occurring psychiatric diagnoses was recorded in 96% of 697 cases,
including 6% only in the clinical record, 9% only in the GAIN and 81% where they agreed. As illustrated
above, “which” diagnosis each found was more likely to vary. About 2/3rds of the time both clinicians and
the GAIN both identified patients with “any” internalizing (54/84) or “any” externalizing disorder (57/93).
With respect to specific diagnoses, the clinicians diagnosed a depressive disorder more often while the
GAIN identified a greater number of the other specific diagnoses.
In Table 2, the first set of columns compares the sensitivity of each method for detecting cases identified by
the clinical expert or self report on the GAIN. The clinicians were more likely to identify any internalizing
disorder and a depressive disorder. Self reports from the GAIN were more likely to identify stress, anxiety
and externalizing disorders. The second set of columns examines the utility for using the GAIN as a
diagnostic screener to predict who is likely to get each diagnosis by a clinician. All are significant at p<.05
and have large positive odds ratio.
Conclusions
. Consistent with earlier research, these data found a high rate of cooccurring Axis-I psychiatric disorders. While there was substantial overall
agreement, there were also many differences - particularly at more
detailed levels of diagnosis. This suggests the value of considering both
clinical diagnosis and self reports from a standardized measure rather
than just relying on one or the other. The results also demonstrate that
that the GAIN was moderately useful as a tool to identify those who are
most likely to benefit from a clinical psychiatric evaluation. The mental
health section of the GAIN can play a vital role, supplemented by other
informants, in the identification of co-occurring psychiatric disorders.
Grant Support
This work was supported by CSAT grant s: TI 11424 (M.Fishman), TI013356 (S.Godley), TI-11432
(P.Shane), TI013313 (T. Gerstel); NIAAA grant AA010368 (M.Godley), NIDA Grant DA 018183
(S.Godley) and CSAT Contract 270-2003-00006 (M.Dennis).