Attempts to treat substance abuse among people with mental health disorders, and vice versa, must begin with recognition.169,41 The purpose of screening is not to determine the complete profile of psychosocial functioning and needs, or to make a diagnosis; but rather to identify whether the individual may have a mental health or substance abuse problem that warrants more comprehensive assessment.
In general, the goal is to have screening instruments that are brief; do not identify a high proportion of false positives; and have good reliability and validity (see Appendix E for definitions of these and other terms related to screening instruments). As noted above the needs and opportunities for identification vary considerably across different types of mental health and substance abuse treatment settings. It is not possible to recommend one approach or screening tool. Therefore, the recommendations are organized into two levels of effort; with the second level requiring more time and expertise than the first, but yielding potentially greater benefits in terms of reliability and validity. Further, this is an area of research that is expanding rapidly and several new screening instruments are available or on the horizon, but which have yet to be validated in either the substance abuse or mental health population. Where appropriate new instruments that are being developed or tested, and which may be appropriate for people with concurrent disorders in either or both populations are mentioned.
In this review the work of Dr. Kate Carey and colleagues40,170,178 has been drawn on for screening. The seminal work of Drs. Robert Drake, Kim Mueser and colleagues has also been valuable.156 The reader is encouraged to review the reports and literature reviews published by these research teams. Several of their reports are listed in Appendix A, and in the References.
The following points concern terminology and some other general issues:
"retell their story". This was an important theme identified in the consumer focus groups;
Within this level four alternatives are described. These alternatives require very little time and effort on the part of clinicians/support workers during the initial contact with the consumer, or within the context of an official intake process. Sensitivity and specificity values derived with a group of people with concurrent disorders are not available for all the alternatives listed. However, such measures and approaches may still have value as part of a clinical decision-making process, especially in settings where more psychometrically sophisticated approaches may not be appropriate during the early period of contact or intake to the program (e.g., acute crisis settings). It is better to cast a wide net in the screening process and subsequently rule out a substance use disorder on the basis of further assessment.
Index of suspicion: If other methods are not feasible or appropriate, it is possible to use a simple checklist of behavioural, clinical and/or social indicators that together can raise the suspicion that the person has a substance use disorder. The following have been considered as common consequences of substance abuse in people with severe mental illness (Group 2). Examples are given in Appendix F.
On the basis of current practice wisdom the following should be added to this list:
Asking a few questions: Some research has shown that the response to a straightforward question about previous problems related to alcohol is highly correlated with the results of more detailed screening instruments.176 The evidence on the value of this approach with people with concurrent disorders is mixed. The reluctance of people to be completely forthright in such self-diagnosis has been noted, and this may be particularly true upon first presentation to some mental health settings where no trusting relationship has yet been established between the consumer and the provider. Indeed, this was identified as a theme in the consumer focus groups. Drake et al.156 warn of the difficulty people with severe mental illness may have in perceiving the relationships between substance use and psychosocial difficulties, and of the tendency to provide socially desirable answers. On the other hand, Barry et al.171 compared consumer self-report and case manager ratings. They found the consumer ratings on some of the questions to be more predictive of a substance use disorder as determined by DSM-III-R criteria. The best predictor of a substance use problem by the consumer was their perception that others were concerned about their substance use (70% sensitivity: 88% specificity; 76% positive predictive value; and 84% negative predictive value). It is cautiously recommended that the three following questions be used as potential Level I screening questions for substance use disorders in mental health settings when other approaches are considered inappropriate. A positive response to any one question should indicate the need for further investigation.
Have you ever had any problems related to your use of alcohol or other drugs? (yes/no)
Has a relative, friend, doctor or other health worker been concerned about your drinking or other drug use or suggested cutting down? (yes/no)
Have you ever said to another person
"No, I don't have [an alcohol or drug] problem, when around the same time, you questioned yourself and FELT,
"Maybe I do have a problem?" (yes/no)
A brief screening instrument: The third approach for Level I screening is based on the CAGE questionnaire,177 and a modification known as CAGE-AID that incorporates both alcohol and other drugs.172 They are considered Level I measures because of their brevity; being comprised of four items which can be routinely incorporated into a formal intake process or a discussion with a consumer seeking help. The CAGE has been validated with a sample of people with severe mental illness and has reasonably high sensitivity and specificity.156,178 Wolford et al.,179 however, compared several screening measures for substance use disorders for people with severe mental illness and, while the CAGE performed better than other approaches such as clinical variables, laboratory tests and collateral reports, it still yielded only modest sensitivity (60.9%) and specificity (69.5%).*The CAGE and CAGE-AID collect information related to lifetime rather than current substance use problems and some may also find this to be a limiting factor.
Case manager judgment: In mental health settings which maintain contact with the consumer for several weeks, months or even years, case managers can ask themselves a few questions to screen for a substance use disorder.171 In the study by Barry et al.171 the best predictor of a consumer's meeting the DSM-III-R criteria for a substance use disorder was the one question:
"Do you think the client has ever had a drinking or other drug problem? Would you say definitely, probably or not at all?"
A relatively new brief screening tool for alcohol use disorders may hold promise for identifying people with concurrent disorders in mental health settings on the basis of further research. This instrument, known as RAPS4, has been developed as a brief screening tool for problematic drinking in emergency room settings.181 It is comprised of four questions related to: Remorse, Amnesia, Performance and Starter (i.e., morning drinking). In emergency settings, a positive response to any one item has been found to have high sensitivity (93%) and specificity (87%). The instrument has also performed well across gender and ethnic sub-groups. Although promising it needs to be validated in mental health settings.
* On some criteria, the TWEAK screening instrument for alcohol problems180 performed better than the CAGE. However, the CAGE has been better researched with people with concurrent disorders and has been adapted to relate to both alcohol and drug problems.
Within this level there are four alternative instruments.* They require somewhat more time and effort to incorporate into routine practice than the Level I alternatives (e.g., there are too many items to commit to memory with a simple mnemonic device such as with the four CAGE questions). However, all measures are still quite brief and easy to administer by interview or self-completion. Also, all the instruments noted below have been validated with people with mental health disorders and they are all in the public domain.
Dartmouth Assessment of Lifestyle Instrument (DALI):173 This instrument is the only screening instrument for substance use disorders that has been developed specifically for use with people with severe mental illness. It consists of 18 items that come from various existing screening tools. It was developed to be interviewer assisted. Eight items predict drug use disorders, nine predict alcohol use disorders. Two items overlap alcohol and drug use disorders. Results suggest it is reliable over time and across interviewers, and that it is more sensitive and specific than several measures including the MAST, TWEAK, CAGE or DAST.173
Michigan Alcoholism Screening Test (MAST):182 Teitelbaum and Carey178 provide a comprehensive review of substance abuse assessment and screening measures applicable for people with severe mental illness. Their review includes many studies including the MAST,182 and its shorter version (SMAST):183 (see also184). The MAST was also one of several screening measures evaluated by Wolford et al.179 While the instrument has been used extensively with people with severe mental illness, it is limited in comparison to the DALI since a separate screening tool will need to be used for drugs other than alcohol** (e.g., the Drug Abuse Screening Test (DAST).185 It also gathers lifetime versus current information. A score of five or more indicates alcoholism; a score of four is suggestive and a score of less than four indicates non-problematic drinking. The SMAST is recommended over the full MAST due to its brevity (12 items). In the recent study by Wolford et al.179 the much shorter self-report scales such as the CAGE or the TWEAK performed equally well as the MAST, if not better. However, all the brief self-report screening tools missed 25% to 40% of the people with alcohol disorders. While the results obtained in other studies with the MAST have been better than found in this recent study (e.g., 86.8% sensitivity noted by Drake et al.156) the MAST or the SMAST need to be complemented by other information such as collateral reports and behavioural observation.
Drug Abuse Screening Test (DAST):185 The DAST is similar to the MAST in that it is based on consumer's self-report and is not diagnostic; being based more on the consequences related to drug use than drug dependence per se. The items can be either interviewer or self-administered. In contrast to the MAST, the DAST items refer to the past 12-months rather than lifetime. Recent research on the DAST with psychiatric outpatient populations has confirmed the internal scale properties with this group and established acceptable test-retest reliability, criterion-related validity, sensitivity and specificity.186,187 In these studies the briefest version of the DAST (10 items) also performed adequately as a screening instrument. The authors of these recent studies recommend a cut-off point of between 2-4 positive items on the DAST-10 as warranting further substance abuse assessment. However, they also point out that different cut-off points can be used depending on the clinician's interest in maximizing sensitivity or specificity. The cut-off point of 2 positive items was reported as achieving a good balance. Maisto et al.187 also point out that the positive predictive value of the DAST-10 was low compared to that reported for the DALI by Rosenberg and colleagues.173 This was attributed to the comparatively low base rate of current drug use disorders in their sample. This underscores the importance of considering the underlying prevalence of substance use disorders in the mental health setting when evaluating the appropriateness of a screening tool. For example, lower prevalence rates will lead to lower predictive value. A tool with low predictive value in a given setting can still be useful if a goal is to limit the number of individuals for whom more extensive, and more costly, assessments of substance use problems would be conducted. The appropriateness of this strategy versus the one of maximizing the number of people screened positive, including false positives, will need to be determined within individual settings and treatment systems.
The Alcohol Use Disorders Identification Test (AUDIT):188 The AUDIT is a well-known, 10-item self-report screening instrument designed to identify people for whom the use of alcohol puts them at risk for negative alcohol-related consequences, or who are experiencing such consequences. Its performance has recently been evaluated with people with severe mental illness.187 The time reference for the AUDIT items is the past year, although a few items have no specific time referent. It can be interviewer or self-administered. Maisto and colleagues187 confirmed the value of the AUDIT in identifying people with alcohol use disorder, or expressing symptoms of that disorder, in the past year. Estimates of sensitivity ranged from .95 to .85 depending on the cut-point used. Specificity ranged from .65 to .77. Consistent with the use of the AUDIT in other settings189 a cut-point of 7 or 8 struck a good balance between sensitivity and specificity when using the diagnostic criteria of DSM-IV as the standard for comparison.
* The project team is aware of the common use of the Substance Abuse Subtle Screening inventory (SASSI) as a screening and assessment tool used by many addictions programs in Canada. The limited validation data for the SASSI generally, and for application with people with concurrent disorders specifically, preclude our recommending it in the present context.
** As with the CAGE instrument a SMAST-AID (i.e. And Including Drugs) has been developed. However, it has not been tested with a sample of people with mental health disorders. Given the potential for confusion in the use of the term
"drug use" the measure can not be recommended for use with this population at this time.
All of the above screening tools are based primarily on consequences related to alcohol or drug use and the item responses do not map onto DSM-IV criteria. One brief tool that is available does provide this mapping and it also covers both alcohol and other drugs with the same set of items. The measure, however, has not yet been extensively evaluated, in particular with people with concurrent disorders. This set of 16 items (Substance Abuse and Dependence Scale: SADS) is a scale within the Global Appraisal of Individual Needs (GAIN:190). The SADS provides a useful screen for dependence (tolerance, withdrawal, inability to control use) and abuse (consequences of use) based on DSM-IV criteria. It also produces a symptom count score which can be used to monitor change over time.
In addition, the Psychiatric Screener described in the next section for screening for mental health disorders, also provides a list of items that map onto the DSM-IV criteria for substance abuse and dependence.
There are many alternatives for screening for substance abuse among people presenting to mental health services. The specific strategy selected may depend on the time and resources available. Asking a few simple questions or using a basic index of suspicion will be better than not giving any attention at all to substance abuse issues. It is also recommended that the results of brief screening tools (e.g., CAGE-AID) be complemented by corroborating information from different sources. Case manager ratings may be particularly helpful in those services with ongoing contact with the consumer. The DALI is the preferred tool for screening for substance abuse among people with severe mental illness.
Best Practice Recommendation
Within this level, there are two alternatives that require very little time and effort on the part of clinicians/therapists during the initial contact with the consumer, or within the context of an official intake process. As with the Level I procedures for screening for substance abuse in mental health settings, reliability, validity, sensitivity and specificity values are not available for these procedures. However, suggestions are based on current practice wisdom and may still have value as part of a clinical decision-making process, especially in settings where more psychometrically sophisticated approaches may not be appropriate during the early period of contact or intake to the program (e.g., withdrawal management settings).
Index of suspicion: If other methods are not feasible or appropriate, it is possible to use a simple checklist of behavioral, clinical and/or social indicators that together can raise the suspicion that the person has a mental health disorder and for whom a subsequent mental health assessment is needed. Consistent with the TIP concurrent disorders protocol,24 the following ABC checklist for a mental health status exam is recommended.
There is a real need for a brief, validated screening instrument for mental health disorders that would be suitable for use in a wide cross-section of substance abuse treatment services. As noted earlier there is also an important distinction to be made between screening instruments that are based on measures/indicators of general psychological distress compared to those with questions that are intended to map directly onto DSM diagnostic criteria. Each approach has advantages and disadvantages. It is also important to keep in mind that the goal of the screening is to identify people who should receive a full mental health assessment at which time diagnosis would be confirmed.
One of the difficulties encountered in identifying potential screening instruments for mental health problems is that the best researched instruments tend not to be in the public domain and therefore require a fee for their use. A good example is the Brief Symptom Inventory which is a 53-item self-report short form of the SCL-90-R.191 It has been used extensively in substance abuse treatment research as a reliable and valid general screen for psychopathology. Another example is the General Health Questionnaire192 and its shorter versions (GHQ-28);193 which has also been widely used in the substance abuse field. There are also brief screening tools specific to some mental health disorders, for example the Centre for Epidemiologic Studies Depression Scale (CES-D):194 and these are cited in the later sections on specific sub-groups of people with concurrent disorders.
Psychiatric Sub-scale of the Addiction Severity Index (ASI):195 The best practice recommendation from among current alternatives in the public domain is the psychiatric sub-scale of the Addiction Severity Index. The scale is comprised of 11 items that tap into previous treatment for psychological or emotional problems; disability pension; use of medication; and experiencing various symptoms (e.g., depression, serious anxiety, hallucinations, cognition difficulties, suicide ideation) but which are not a direct result of drug/alcohol use. In addition to these 11 items, both the client and the therapist provide various ratings of problem severity. Through communication with the developers of the ASI, the following four questions can supplement the ASI Psychiatric Sub-scale in its published form.
Have you had a significant period of time (that was not a direct result of alcohol/drug use) in which you have (0=no; 1=yes):
There are three promising screening tools under development:*
There are alternatives available for screening for mental health disorders among people presenting to substance abuse services. The specific strategy selected may depend on the time and resources available. Asking a few questions or using a checklist for mental health status will be better than not giving any attention at all to mental health issues. The psychiatric subscale of the Addiction Severity Index is recommended and should be supplemented by a small number of additional items. Promising new screening tools for mental health disorders are currently under development.
Best Practice Recommendation
* Another measure known as PRIME-MD screens for psychiatric disorders but has been developed spcecifically for physicians.196 While it may have some potential value in treatment settings with a staff physician it would require some modification to be more widely applicable. Further the tool has not been tested and validated with people with concurrent disorders.
** The contact person for information about this screening tool is Mr. Wayne Skinner (1-416-525-8501, Ext. 6387)