Retrospective self-report
Self-monitoring
Self-report screening tests for substance use disorders
Biochemical assays
Collateral information
Screening for mental disorders in people with substance use disorders
Assessment of psychiatric symptoms in people with substance use disorders
Assessment of insight
Assessment strategies covering both mental disorders and substance use disorders
Distinguishing true comorbidity from secondary effects
Assessment of readiness to change
Summary and recommendations

Retrospective self-report

Given the high prevalence of comorbidity, especially in treated populations, inquiries about each disorder should routinely be undertaken when the other is detected. Failure to do this will result in a significant proportion of people with comorbidity being missed (Appleby et al., 1997). Sound assessment requires the prior development of rapport with the patient, so that the person feels safe to disclose substance use without fear of rejection or other punitive responses. This general principle is especially important when the person is currently paranoid or is being assessed for involuntary treatment. In people experiencing positive psychotic symptoms, selective attention and other cognitive deficits can present further problems — minimising distraction, using simple sentence construction and repeating questions may assist in maintaining attention. Self-reports that are gathered in sub-optimal conditions may need to be checked against later reports, and either collateral information or biochemical data to examine their accuracy.

Questioning about usual intake tends to result in an underestimation, especially in heavy users (Feunekes, van 't Veer, van Staveren, & Kok, 1999; Townshend & Duka, 2002). There are several reasons for this problem. For example, where there is substantial variability of intake from day to day (e.g., based on substance availability), judgments about usual intake are prone to retrospective biases based on beliefs about usual intake and the salience of consumption occasions (Nisbett & Ross, 1980). Respondents can provide an index of the difficulty they experienced in providing a "usual" quantity, frequency estimate as a result of this variability (Hasin & Carpenter, 1998). However the problem can be better addressed in the interview setting by use of the Timeline Followback method (Sobell & Sobell, 1995), which uses situations and events that the person has experienced to cue recall of consumption. This strategy allows the gathering of accurate retrospective data on consumption over recent weeks or months. The method is of course difficult to undertake when the person is acutely thought disordered. If there is a risk of underestimation, examples of high levels of consumption may reduce unwillingness to report these and shift the reporting anchor (Nisbett & Ross, 1980). In the case of alcohol, additional issues include the difficulty in people summating intake of different types of drinks (Feunekes et al., 1999, suggesting a need for separate questioning on each type), and a substantial underestimation of amounts poured at home (Kaskutas & Graves, 2000), emphasising the need for concrete examples (and often, measurement practice) when explaining the size of standard drinks.

The Opiate Treatment Index (OTI, Darke,Ward, Hall, Heather, & Wodak, 1991) is a structured interview which assesses demographic characteristics and treatment history, consumption of 11 classes of drugs during the month preceding interview, as well as HIV risk-taking behaviour, health, social functioning, criminality and psychological comorbidity (General Health Questionnaire GHQ-28, Goldberg & Williams, 1988). The OTI has excellent psychometric properties (Darke, Hall, Wodak, Heather, & Ward, 1992). Baker and colleagues (in press) have used this instrument successfully with people with severe mental disorders on in-patient and outpatient bases. However, as clinical trials with the OTI indicated that there were difficulties with the recent use episodes methodology, Teesson, Gallagher and Ozols (1997) modified the OTI for use among people with severe mental disorders by expanding the treatment history section to include psychiatric history; referring to drug use over the preceding three months; adding cough syrup and medications for side-effects as two additional drug classes; and eliminating a question concerning conflict with relatives.Top of page

Self-monitoring

Daily monitoring of substance use can provide excellent data about intake, and variants can also give information about the situations in which consumption is most likely to occur, the cognitions or activities that precede it or the effects that follow. The accuracy is of course limited by delays before recording and by effects of intoxication upon memory. It can be difficult in some social settings to take out a monitoring form and record use at the time, but surreptitious recording strategies (e.g., moving a coin from one pocket to another) and noticing the products of use (e.g., empty bottles, cigarette ends) can aid later recall. Self-monitoring may also have reactive effects on intake (Kavanagh, Sitharthan, Spilsbury, & Vignaendra, 1999), especially where the person records their consumption against a contracted intake goal. However as in other contexts, systematic completion of the monitoring on a daily basis can be an onerous task in itself, and adherence to the self-monitoring is a significant clinical challenge. In comorbid populations, there is little data on adherence with self-monitoring, but in severe mental disorders it may prove to be especially difficult to obtain daily data. Devices to remind the person to record self-monitoring data have been successfully used in smoking research (Shiffman et al., 1997), and may assist in the collection of data in comorbid populations as well.

Self-report screening tests for substance use disorders

Screening for significant substance abuse or dependence in mild mental disorders can usefully apply standard screening tests (Dawe, Loxton, Hides, Kavanagh, & Mattick, in press). However in more severe disorders, where only a small proportion of comorbid patients have high levels of physical dependence (D. J. Kavanagh et al., 1999), some of these measures are insufficiently sensitive to detect substance abuse, and others require a high degree of intact cognitive functioning that may not be present. Thus, measures such as the Addiction Severity Index (ASI), (McLellan, Luborsky, Woody, & O'Brien, 1980), Michigan Alcoholism Screening Test (MAST), (Selzer, 1971) and the CAGE alcohol questions (Ewing, 1984) perform relatively poorly in severe mental disorders (Carey, Cocco, & Correia, 1997; Wolford et al., 1999; Zanis, McLellan, & Corse, 1997). In contrast, the Alcohol Use Disorders Identification Test (AUDIT, Saunders, Aasland, Babor, Fuente, & Grant, 1993), which is a sensitive measure of both milder and more severe forms of alcohol misuse in the general population, is also appropriate for use in populations with severe mental illness (D. J. Kavanagh et al., 1999; Maisto, Carey, Carey, Gordon, & Gleason, 2000; Seinen, Dawe, Kavanagh, & Bahr, 2000). The Severity of Dependence Scale (SDS, D. J. Kavanagh et al., 1999) also performs well in identifying substance use disorders in those with severe mental illness.

Some screening measures have been especially designed for this population. In the US context, the Dartmouth Assessment of Lifestyle Instrument (DALI), (Rosenberg et al., 1998) has demonstrated high levels of sensitivity and specificity to alcohol, cannabis or cocaine abuse within in-patients with severe mental disorders. Locally, the DrugCheck Problem List (D. J. Kavanagh et al., 1999) has also shown a high rate of correct classification in relation to full interview assessment.

Biochemical assays

Standard biochemical assays for substance use can assist in validating self-reports. However the accuracy of these assays is limited by the duration over which the substance and its metabolites may be accurately detected and by the detection levels that are set in the analyses. Breath analysis gives a particularly short window for detection of alcohol or cigarette use (e.g., carbon monoxide from cigarette smoke can be reliably detected over about 6 hours). In most cases, urine samples will detect the parent drug or its metabolites within 48 hours of last use of the drug. Sometimes an associated substance may be detected by biochemical measures over a longer period — for example, salivary thiocyanate (reflecting cyanide present in cigarette smoke) has a half-life of about 9.5 days. Hair samples allow analysis of substance use over several weeks by capture of the substance within the growing hair, and this method is readily used even in severe mental disorders (McPhillips et al., 1997). However hair analysis has not as yet become a standard clinical procedure.Top of page

Collateral information

Data from other informants may also be used for validation of the self-reports, although collateral informants are themselves prone to biased reporting, including minimising or exaggerating current use, or being affected by beliefs about the person and their intake (e.g., that their substance use will not change). They are also likely to be influenced by salient past events (e.g., unusually high levels of use or the theft of personal items, Nisbett & Ross, 1980) and misattribution of symptoms (mistaking symptoms of mental illness for evidence of substance use and vice versa). In addition, collaterals may not have full information about the substance use, especially if the substance use is being concealed. They may be more likely to accurately report substance use when they are in close contact with the person (Carey & Simons, 2000).

Biochemical assays and collateral data not only attest to the validity of self-reports, they can also encourage accuracy in self-reports when the person is aware of the check being in place (the 'bogus pipeline effect', Aguinis, Pierce, & Quigley, 1995). In practice however, self-reports are usually reliable as long as incentives for accurate reporting are in place. The addition of biochemical assays or collateral assessments does not usually add substantially to this accuracy (Rankin, 1990), even in comorbid substance use and severe mental disorders (Carey & Simons, 2000). In an urban US sample of psychiatric patients attending an emergency department, a self-reported history of intake over the previous three days was more likely to result in reported use of alcohol and cannabis than a urine screen (Perrone, De Roos, Jayaraman, & Hollander, 2001). The urine screen did not show significantly more identified cases than the history on any substance.

Screening for mental disorders in people with substance use disorders

Screening for mental disorders in people with substance use disorders may use standard tests that have been developed for the general population (Dawe et al., in press). The General Health Questionnaire (GHQ), (Goldberg & Williams, 1988) is a self-administered questionnaire that provides a measure of generalised distress, with the 12-item version performing about as well in detecting disorder as the 28-item version (Goldberg, Gater, Sartorius, Ustun, & et al., 1997). An alternative to the GHQ is the Self-Reporting Questionnaire (SRQ), (Beusenberg & Orley, 1994), which has been validated internationally as a screen for psychiatric disorder. Versions have been examined with 20 or 24 items. The Symptom Check List-90-Revised (SCL-90-R), (Derogatis, 1994) — also a self-report measure — displays high sensitivity and moderate specificity for anxiety and depressive disorders in patients with substance misuse, and is better able to identify these disorders than the Addiction Severity Index (Franken & Hendriks, 1999). It is only available for use under the supervision of a clinical psychologist. Brief forms of the scale such as the Brief Symptom Inventory (BSI, Derogatis & Meilisaratos, 1983) show high correlations with the full SCL-90-R.

The Brief Psychiatric Rating Scale (BPRS), (Lukoff, Liberman, & Nuechterlein, 1986; Overall & Gorham, 1962) is a clinician-completed scale providing both screening and detection of changes in symptoms. It has been commonly used in studies on comorbid populations as a criterion measure (e.g., Dixon, Haas, Weiden, Sweeney, & Frances, 1991; Warner et al., 1994). The scale takes about 20 minutes to complete, and includes both standard questions and observational ratings. Detailed descriptions have been published (Ventura, Green, Shaner, & Liberman, 1993; Woerner, Mannuzza, & Kane, 1988) to assist in anchoring ratings. The Positive and Negative Syndrome Scale (PANSS, Kay, Fiszbein, & Opler, 1987) is an adaptation of the BPRS that was designed to provide scores on positive and negative syndromes of schizophrenia and general psychopathology. Both scales require training and calibration of interviewers to ensure reliability and validity of the assessments.

Assessment of psychiatric symptoms in people with substance use disorders

The Psychiatric Diagnostic Screening Questionnaire (PDSQ), (M. Zimmerman & J.I. Mattia, 2001; M. Zimmerman & J. I. Mattia, 2001) is a recently developed self-report screening instrument that requires approximately 20 minutes to complete and produces predictions for a broad range of 13 common DSM-IV disorders, including alcohol and drug use disorders, as well as major depression, bipolar disorder, post-traumatic stress disorder (PTSD), and psychosis. Studies on the PDSQ have indicated good test-retest reliability, and high sensitivity, specificity, and predictive value when compared with structured clinical interviews. Considering the ease of administration, the strong association with structured clinical interviews, and experience with the scale reported in several thousand people, the PDSQ would appear to have broad applicability in mental health, substance abuse, or primary health care settings.Top of page

Assessment of insight

Insight is one of the most consistently reported predictors of compliance with psychiatric treatment. The Schedule for the Assessment of Insight (SAI), (David, 1990) is a semi-structured interview that measures three aspects of insight: willingness to accept that one has an illness; ability to correctly label psychotic experiences; and acceptance of treatment. Scores on this scale are expressed as a percentage of maximum insight. Kemp and colleagues (1998) reported a significant difference in scores on this measure as a function of compliance therapy, based on motivational interviewing and cognitive approaches to psychotic symptoms and supportive counselling among 74 acutely psychotic in-patients. The Insight Scale for Psychosis, (Birchwood et al., 1994) was shown to have adequate psychometric properties among 133 subjects with varying non-affective psychoses and is suggested as a quick and acceptable self-report measure that is reliable, valid, and sensitive to individual difference and change.

Readiness to change substance use or to accept treatment does not necessarily require full insight. If efforts are concentrated on agreement with a diagnosis or even a full understanding of symptoms, a failure to obtain that agreement often becomes a block to engagement. Furthermore, the acceptance of disorder often produces dysphoria and a loss of self-efficacy that can damage commitment to change. Understanding can initially be partial, as long as it is sufficient to motivate some change. Assistance can then focus on the issues that are effectively motivating the person.

Assessment strategies covering both mental disorders and substance use disorders

A number of standardised diagnostic interviews are available to assess both substance use disorders and other mental disorders in a single assessment. These include the Composite International Diagnostic Interview (CIDI, Semler et al., 1987), the Structured Clinical Interview for DSM-IV (SCID, First, Spitzer, Gibbon, Williams, & Benjamin, 1994; Spitzer, Williams, Gibbon, & First, 1992), and the Schedules for Clinical Assessment in Neuropsychiatry (SCAN, Wing et al., 1990). The full interviews take some time to administer and score, although segments can be selected for specific focus (e.g., substance use, anxiety disorders). A personality disorders segment of SCID is also available (Structured Clinical Interview for DSM-IV Axis II Personality Disorders—SCID-II, First et al., 1994). All of the structured interviews require training and calibration on the conduct of the interviews and the use of interviewer ratings. The CIDI is available in a computerised, self-administered version (CIDI-Auto), (Peters & Andrews, 1995), and comparable results on anxiety and depression are available in that format (Peters, Clarke, & Carroll, 1999). Some recent data suggests that there may be problems with the concordance of diagnoses derived from CIDI and SCAN interviews given sequentially (Brugha, Jenkins, Taub, Meltzer, & Bebbington, 2001).

The Primary Care Evaluation of Mental Disorders (PRIME-MD), (Spitzer et al., 1994) is a questionnaire and interview instrument designed for detection of psychiatric or substance use disorders within primary care settings. A version of the interview can be administered over the telephone using interactive voice response technology (Kobak et al., 1997). PRIME-MD found high sensitivity and specificity in relation to the Structured Clinical Interview for DSM-IV (SCID). The computerassisted version of PRIME provides similar results to the face-to-face interview, except that more patients report alcohol-related problems on the computerised version (Kobak et al., 1997).

Cognitive dysfunction is an issue in both mental illness and substance use disorders. The Mini-Mental State Examination (MMSE), (Folstein, Folstein, & McHugo, 1975) is a helpful quick screen for cognitive dysfunction that does not require psychological training (Mattick & Jarvis, 1993).Testing should not be conducted during detoxification and only when the client is sober Mattick & Jarvis (1993); and Saunders & Robinson, (in press) recommended that a small cadre of staff on each alcohol and other drug agency be trained in the MMSE as it assists in recognising the presence of mental health symptoms and improves ability to communicate with mental health staff.Top of page

Distinguishing true comorbidity from secondary effects

At initial presentations, it is often difficult to distinguish between effects of substance use and psychiatric symptoms. For example, symptoms of anxiety or depression can arise during intoxication or withdrawal from a variety of substances. Similarly, there are few differences between the acute symptoms of schizophrenia and those from a transient substance-induced psychosis. Some diagnostic issues may be clarified by taking a careful history from the patient or other informants and by tracking the resolution of symptoms. For example, a stimulant-induced psychosis usually resolves within days of ceasing the stimulants (Schuckit, 2000). However the basis of current symptoms can often be unclear in cases where both problems have been present at times in the past.

In some cases the basis of a specific symptom may be of little immediate importance for its management. The immediate pharmacological treatment may often be identical, although often shorter in duration when psychiatric symptoms are primarily due to substance use. Where both substance use and mental disorders are current, the psychological management may often need to address both disorders rather than be confined to the trigger for symptoms on this occasion (K. T. Mueser et al., 1998). Detecting current causal influences does have importance even in cases where much of the treatment is unchanged, since it allows better prediction of outcomes (e.g., if increased symptoms have repeatedly been associated with greater substance use, this may well recur at the next exacerbation). However the identification of the causal influences is often difficult or impossible to disentangle, and delaying treatment of one or both disorders in order to determine one's "primacy" or the instigating trigger/s can have deleterious effects on both disorders. In consequence, the detection of causal influences is generally conducted in the context of attempting to treat both disorders.

Assessment of readiness to change

Prochaska, DiClemente and colleagues (e.g., Prochaska et al., 1986; Prochaska et al., 1992) have proposed a transtheoretical model that can be used to assess readiness to change and allow appropriate tailoring of interventions. During 'pre-contemplation' the person is not considering change, and is either unaware of the benefits of change or is contented with their behaviour. At 'contemplation', the person is ambivalent about change but has not determined to change. During 'preparation', the person prepares to take action. The 'action' stage is characterised by active attempts to change. During the 'maintenance' stage the person focuses on maintaining the changes made. People who are in earlier stages of change tend to respond to action-oriented strategies with resistance. Motivational interviewing may help to prepare the person for change, nudging them from pre-contemplation to contemplation and preparation. It can also be employed to encourage optimism and self-efficacy in the action and maintenance stages of change (Miller & Rollnick, 2002).

How a person with comorbid psychosis and substance misuse fares in treatment depends on their readiness to acknowledge both disorders (Smyth, 1996), adhere to medication (Kemp, Hayward, Applewhaite, Everitt, & David, 1996; Kemp, Hayward, Applewhaite, Everitt, & David, 1998), and participate in non-pharmacological interventions such as vocational rehabilitation programs (Rogers et al., 2001). Their readiness to change in each of these areas should be assessed. For example, insight into illness will affect recognition of the role of medication and the potential for substance use to exacerbate symptoms and interfere with effectiveness of medication. Stage of change should be assessed early on in treatment and reassessed regularly, providing opportunities for early detection and treatment matching (Ziedonis & Trudeau, 1997). Ongoing stage of change assessment over the longer-term is also important because participation in treatment for mental or substance use disorders, especially when coerced, may lead to a return to previous behaviour when pressure to comply has been lifted (Smyth, 1996).Top of page

Readiness to change alcohol and other drug use

Although the utility of self-report measures of readiness to change substance use among people with psychosis and substance use disorders has been questioned (Addington, el-Guebaly, Duchak, & Hodgins, 1999), others have used such measures successfully. Oral administration of questionnaires may be helpful (e.g., Carey, Purnine, Maisto, & Carey, 2001). Velasquez, Carbonari and DiClemente (1999) measured stages of change, decisional balance, temptation and self-efficacy among 132 alcohol dependent people in a public mental health clinic's outpatient dual diagnosis program. Primary diagnoses for the sample comprised depression (41%), schizophrenia (30%), bipolar disorder (16%), psychosis (7%), mood disorder (3%), anxiety (2%) and adjustment disorder (1%). The Readiness to Change Scale of the 28-item University of Rhode Island Change Assessment Scale-Alcohol (URICA-A, DiClemente & Hughes, 1990) was used to assess readiness to change drinking. Cronbach's alpha was 0.91 for the URICA measure, indicating that presence of an Axis 1 mental disorder may not be associated with poor internal consistency of instruments designed to measure readiness to change substance use, although test-retest reliability was not assessed (Carey et al., 2001).

Carey and colleagues (2001) subsequently evaluated the psychometric adequacy of three instruments designed to assess readiness to change substance misuse among 84 people with severe and persistent mental illness. The instruments were the Stages of Change Readiness and Treatment Eagerness Scale (SOCRATES, Miller & Tonigan, 1996), the Decisional Balance Scale (DBS), (King & DiClemente, 1993), and the Alcohol and Drug Consequences Questionnaire (ADCQ, Cunningham, Sobell, Gavin, Sobell, & Breslin, 1997). All three instruments demonstrated good internal consistency, reliability and validity. Carey and colleagues concluded that the use of self-report instruments of readiness to change can be justified among people with severe mental illness and substance use disorders, and that such measures may also be usefully included in outcome assessments.

The 12-item Readiness to Change Questionnaire (RTC), (Rollnick, Heather, Gold, & Hall, 1992) has been employed in several studies among people with co-occurring psychiatric and substance use disorders (e.g., Blume & Marlatt, 2000; Blume & Schmaling, 1997; Blume, Schmaling, & Marlatt, 2001; Claus, Mannen, & Schicht, 1999). This questionnaire yields three scores, Pre-contemplation, Contemplation and Action, and a total score that provides a measure of overall motivation to change. It has been found to have satisfactory internal consistency and test-retest reliability (Rollnick et al., 1992) and to have predictive validity for drinking rates among people without comorbid problems (Heather, Rollnick, & Bell, 1993).

Some brief measures of the stages of change for substance use appear to be reliable and valid in comorbid populations, particularly when they are orally administered. However, further research in this area is required to establish the psychometric properties of these instruments among people with comorbid disorders.

Readiness for treatment and treatment adherence

There have been few studies examining readiness for treatment among people with comorbid mental and substance use disorders. The Substance Abuse Treatment Scale (SATS), (McHugo, Drake, Burton, & Ackerson, 1995) is a clinician-rated scale assessing stage of change of substance abuse treatment during the past six months among people with severe mental illness and SUD. Psychometric data were derived from seven community mental health centres over a three-year period from clients, case managers, families and mental health and non-mental health treatment and service providers. The SATS has adequate content, construct and criterion validity and adequate test-retest and inter-rater reliability (Teesson, Clement, Copeland, Conroy, & Reid, 2000).

Rogers and colleagues (2001) assessed readiness for treatment using the Change Assessment Scale (CAS), (McConnaughy, Prochaska, & Velicer, 1983), a 32-item questionnaire, among 163 comorbid individuals, primarily with psychotic disorder, major depression or bipolar disorder. The instrument was administered at baseline and 24 months after entry to the study. Participants were asked to consider their vocational or employment situation as they completed the items. Rogers et al. (2001) reported that the contemplation, action and maintenance sub-scales achieved the satisfactory level of internal consistency found in the non-comorbid original sample. However, the pre-contemplation scale did not achieve this level of internal consistency and did not correlate negatively with the maintenance sub-scale.The authors suggested that people with comorbid disorders may be less aware of their need to change and remain more entrenched at the pre-contemplation stage. Factor analysis revealed considerable overlap between the contemplation and action stages. They suggested that past work failures may have led to more ambivalent attitudes about attempting vocational changes.The CAS was able to predict early attrition in that there were significant differences between dropouts and completers. However, the CAS was not able to predict later behaviour change well and the authors suggested that readiness to change may not be a stable phenomenon, making it difficult to use as a predictor of long-term change. The authors suggested that further studies of the instrument's psychometric properties be undertaken and suggested it may be a reasonable predictor of proximal rather than long-term change.Top of page

DeLeon (2001) has suggested that as client motivation and readiness for treatment have been shown to be important in treatment retention, brief, reliable, valid and user-friendly questionnaires should be used to assess these areas. He suggested using the Texas Christian University scales (Joe, Simpson & Broome, 1998) and/or the Circumstance, Motivation, and Readiness (CMR) scales (DeLeon, Melnick, Kressel, & Jainchill, 1994), measuring external motivation, internal motivation and readiness for treatment. No data was reported on their use among people with comorbid mental and substance use disorders.

The Attitudes to Medication Questionnaire (ATM) is a 14-item semi-structured interview designed to measure patients' attitudes to psychotropic drugs, developed by Hayward et al. (1995). Patients are asked about their feelings about their medication, the role of staff in dispensing their medication, and their plans for the future. Higher scores indicate more positive attitudes towards medication. The test-retest reliability of this measure was 0.77 in a small pilot study. Kemp et al. (1998) have reported a treatment effect for scores on this measure.

Compliance with medication is difficult to measure. An indirect measure of compliance used by Kemp et al. was sensitive to differences between control and treatment groups. This measure correlated highly with attitudes to medication. Estimates of compliance were rated on a Likert scale (1=complete refusal, 7=active participation) by at least two people involved in the care of participants (eg., health practitioners, relatives) and converted into a composite compliance score. The Medication Adherence Rating Scale (MARS), (Thompson, Kulkarni, & Sergejew, 2000), a refinement of the Drug Attitude Inventory (DAI), (Hogan, Awad, & Eastwood, 1983), was administered to 66 people, the majority diagnosed with schizophrenia. Lithium levels and carer ratings were also recorded to verify compliance when available. Results indicated that the inventory appeared to be a reliable and valid measure of compliance to psychoactive medications.

Importance, confidence and readiness

Rollnick, Mason and Butler (1999) have described the use of open-ended questions to assess importance, confidence and readiness to change. They suggest that the following line of questioning may be helpful: "How do you feel at the moment about [change]? How important is it to you personally to [change]? If zero was 'not important' and 10 was 'very important', what number would you give yourself?" [p 63]. Similar questions can be asked of confidence about behaviour change and overall readiness to change. Rollnick et al. recommend that motivational interviewing may be indicated when a person indicates low importance, whilst high importance and low confidence may indicate training the person in strategies to enhance confidence and ability to change. Assessment of importance, confidence and readiness in this way may be an efficient and non-intrusive tool for the concurrent assessment of stage of change for mental and substance use disorders; adherence to medication; and participation in non-pharmacological interventions. No evidence on the specific utility of this assessment in comorbid populations is available at present.

Summary and recommendations

The selection of assessment instruments will depend on the context and the assessment objectives. Standard screening instruments for substance use disorders and for mental disorders should routinely be used in situations where staffing time or expertise prohibit the universal application of more extended assessments. Without this routine screening, cases of comorbidity will be missed. Procedures also need to be in place to alert staff to conduct additional assessment for comorbidity in positively screened cases. The AUDIT for alcohol and the SDS, DALI and DrugCheck for other drugs appear to be performing well as screening instruments. On current evidence we have no reason to doubt the validity of standard instruments such as BPRS for the assessment of psychiatric symptoms within people with substance use disorders, although further data specifically addressing their use by substance treatment staff is required. Current data on PRIME-MD offers effective mental disorder and substance use disorder screening in primary care, and the voice-activated telephone method may provide significant cost advantages over a live interview approach.Top of page

Ideally, a standardised interview should be used to validate diagnoses, and this will be especially important in research. With appropriate training, the SCID and SCAN perform particularly well. The computerised CIDI may also assist with diagnosis, but current data suggests that such assessment should be supplemented by observation and checking of responses by trained and experienced clinicians.

Retrospective self-reports of substance intake appear to provide valid estimates when there are incentives for accuracy and where a Timeline FollowBack technique is used to assist recall. Knowledge that collateral information and/or biochemical assays are being collected may assist in maximising accuracy. The accuracy of collateral reports requires observational access and is subject to the same potential retrospectivity and biasing constraints as self-report. The utility of biochemical assays is subject to the speed of elimination of the substance and its metabolites and to cost constraints, but should be considered a standard procedure at in-patient admission and at random intervals during treatment. Self-monitoring of substances and symptoms presents challenges in people with significant cognitive impairment or high negative symptoms, but variants such as brief telephone interviews can assist. Assessments of insight such as the SAI or ISP, and of mental status using the MMSE provide important additional information.

It appears that readiness to change substance use may be reliably assessed with measures such as the RTC, SOCRATES, DBS and ADCQ, although further data is needed. Further work on the development of assessments of readiness for treatment within comorbid populations is required before other specific recommendations can be made. Assessment of readiness to change may present particular challenges with involuntary patients, where there is a risk of over-reporting intentions in order to obtain desired changes in treatment status.

This field is developing very quickly. Over the next few years, we can expect a variety of psychometrically sound assessment instruments and protocols to emerge for use in specific contexts. Cost-effective administration procedures including computer-based methods can also be expected to develop further, although in comorbid populations with cognitive deficits or limited insight we may anticipate that live interview assessment will still be needed.

Accuracy of assessment relies on the development of rapport and trust, and this will be one reason why we expect that live assessment will continue to have an important role. As in other populations, assessment and treatment necessarily come together in the development of this rapport. Aspects of the following section are therefore also of key importance in the assessment process.