2.9.1 Instruments and outcome measures

Assessments by independent research-assistants
Extensive baseline and follow-up data were collected in the study, both to screen potential study candidates during the qualification phase, to provide an extensive baseline description of the study population, and to investigate pre-treatment to post-treatment changes in the physical, psychiatric, social, and substance use status of the participants. The Addiction Severity Index (ASI; McLellan et al., 1980, 1992; Hendriks et al., 1989) was used as the basic instrument in the study, complemented with various supplements in areas of functioning that required more extensive data. In the present study, the European version of the ASI was used (Blanken et al., 1994; Kokkevi and Hartgers, 1995). The ASI, for which an extensively validated Dutch translation is available, is a semi-structured interview which assesses the status and severity of problems of subjects in the areas of physical health, employment, alcohol and drug use, legal functioning, social functioning, and psychiatric functioning. In previous studies, the ASI has shown sufficiently high inter-rater reliability, test-retest reliability, and concurrent and discriminant validity across a range of alcohol and drug dependent populations (McLellan et al., 1985, 1992; Hendriks et al., 1989; Kokkevi and Hartgers, 1995; Alterman et al., 2000). For the purpose of the present study, the standard items of the ASI in the area of social functioning and social integration were supplemented with detailed questions about current illegal activities, income, and living arrangements of the patient, and an item was added which informed about the number of days the subject had had personal contact with non-drug-using persons in the previous month.
In the present trials, the mean number of days in the previous month the subject had been involved in illegal activities amounted to 10.9 (sd=11.4), and the median to 8. The mean number of days without contact with a non-drug-using person amounted to 14.7 (sd=12.8), with a median of 12.

In the area of physical health, the Maudsley Addiction Profile Health Symptoms Scale (MAP-HSS; Marsden et al., 1998) was used as a supplement to the ASI. The MAP-HSS is a ten-item structured interview, which was adapted from the health scale of the Opiate Treatment Index (Darke et al., 1991, 1992). Each item is scored on a five-point Likert-type scale, ranging from 0 (complaint never present in the previous 30 days) to 4 (complaint always present in the previous 30 days), resulting in a total scale-score ranging from 0 to 40. In previous research among opiate users in opioid substitution, detoxification, and relapse prevention treatment in the United Kingdom, the scores on the MAP-HSS were approximately normally distributed, the internal consistency of the scale was satisfactory ( =0.79), and the test-retest reliability of the scale was high, with an intra-class correlation coefficient amounting to 0.86 (Marsden et al., 1998).
As described in paragraph 2.1.2, patients had to have a MAP-HSS score of at least 8 to be included in the present trials in the area of physical health. This inclusion threshold score was based on the instrument's distribution in the study of Marsden et al. (1998), and was determined by subtracting one standard deviation of the MAP-HSS total score (sd=6.9) from the mean total score (mean=14.7) found in this sample.
In the context of the present study, the psychometric properties of the MAP-HSS were investigated in a sub-sample (n=808) of the Dutch general population, matched on age and gender with the study population of the heroin trials. In the general population, the mean MAP-HSS score amounted to 4.8 (sd=4.7), and the median to 4. Utilizing the threshold score of at least 8 on the MAP-HSS, 22% of the males and 25% of the females in the general population in the Netherlands met the threshold. Lastly, the internal consistency of the MAP-HSS in the general population amounted to =0.80.
In the study population of the present trials the mean MAP-HSS score was 11.4 (sd=7.4), and the median 11. Cronbach's of the scale in the study population amounted to 0.82.

In the area of psychiatric status, the ASI was supplemented with the Composite International Diagnostic Interview (CIDI; Robins et al., 1988; WHO, 1996) and the Symptom Checklist (SCL-90; Derogatis, 1983; Arrindell and Ettema, 1986). Like the ASI, the CIDI and SCL-90 are international standards, for which validated Dutch translations are available. In the present study, the computerized version of the CIDI was used to obtain DSM-IV diagnoses (APA, 1994). The SCL-90 is a self-report questionnaire consisting of 90 items, which are scored on a five-point Likert scale, ranging from 0 (symptom absent) to 4 (symptom very often present). Hence, the total score of the SCL-90 could range from 0 to 360. The psychometric properties of the SCL-90 have been established in the general population, and across a wide range of clinical populations, including alcohol and drug dependent patients (Arrindell and Ettema, 1986). In Dutch samples of treatment seeking alcohol and drug addicts, the SCL-90 scales showed high internal consistency, with coefficients ranging from 0.81-0.89 (Hendriks, 1990a, 1990b), and a clear sensitivity to change from pre- to post-detoxification treatment (Franken and Hendriks, 2001).
As described in paragraph 2.1.2, the cut-off score on the SCL-90 to be included in the trials in the area of mental health was 41 for males and 60 for females. These inclusion threshold scores were based on the distribution of the SCL-90 total score in a general Dutch population norm group (Arrindel and Ettema, 1986). Since the SCL-90 total score showed a skewed distribution in both the male and female norm group (males: mean=27.2; sd=27.3; females: mean=38.9; sd=36.4), the cut-off score was derived from the minimum value of the total scores falling within the 80th percentile in the norm group (i.e. the percentile in which 20% of the general population scores "high" to "very high" on the SCL-90). For males, this minimum value amounted to 41, and for females to 60.
In the study population of the present trials, the mean SCL-90 score amounted to 67.7 (sd=56.2; median=51) for males, and to 95.8 (sd=76.0; median=72) for females. The internal consistency of the SCL-90 total scale in the study population amounted to =0.98.

In addition to these standardized instruments, two supplements - both structured questionnaires, which were specifically developed for the present study - were added to the CRF. One supplement was directed at obtaining detailed information about the type and amount of additional (somatic, psychiatric, social, and substance abuse) treatment the patient had received during the study. The other supplement informed about the patient's satisfaction with regard to the treatment provided in the treatment site. Lastly, urine samples were collected and analyzed on various substances by an independent central laboratory, and registered data in the police records were investigated by an independent research-assistant, to verify the self-report data in the areas of substance use and criminality, respectively (see paragraph 3.2.8).

To minimize information bias, the study assessments were conducted by independent research-assistants, who used standardized instruments and evaluation procedures, and the results of the assessments (including those from the urinalyses and the police records) were not made available to the treatment staff. To obtain standardized and reliable data, all research-assistants received extensive training courses and Go back or Close Windowster sessions in the administration of the questionnaires and interviews. In addition, comprehensive Dutch manuals for the assessment and scoring procedures were available for each instrument and supplement.

Assessments by the treating physician
The instruments and supplements described above were all intended to collect information, directly related to the eligibility of potential study candidates and the effectiveness of the interventions in the study, and were therefore administered by independent research-assistants. In addition, the CRF contained various supplements which did not directly pertain to the effectiveness of the intervention. These supplements were filled out by the physician to document the findings of medical examinations (including the results of laboratory blood tests and pregnancy tests), the amount and type of prescribed co-medication, the physician's rating on a GAF-scale with regard to the patient's level of psychosocial functioning, and the occurrence of (serious) adverse events (see paragraph 2.10.1). Like the research-assistants, all physicians were extensively informed about the assessment procedures involved in using these supplements. Table 4 provides an overview of the measures collected in the areas of physical and mental health, social functioning and integration, and substance use in the study.

Table 4. Study measures