Injection Drug Use Quality of Life scale (IDUQOL): A validation study

Participants consisted of a sub-sample of individuals participating in the Vancouver Injection Drug User Study (VIDUS), a longitudinal study of the incidence of HIV among IDUs in Vancouver, Canada. The research design and methods of the VIDUS have been previously described [17]. In brief, this open cohort study was initiated in 1996 to clarify the socio-demographic and behavioural determinants of HIV sero-conversion among this group. Eligibility for initial enrolment required current injection drug use (injected at least once within the last month) and evidence of recent injection was required by inspection of needle tracks. Potential participants also were required to reside in the Lower Mainland of British Columbia and provide informed consent. Most participants (82%) were recruited through word of mouth and street outreach programs. The remaining participants were referred by the needle-exchange program (5%), other storefront agencies (10%), and clinics (3%). Participants who have stopped injecting after the baseline visit are still eligible for follow-up. Trained interviewers administer a survey instrument every 6 months. Participants are asked about their demographics, needle sharing, drug using behaviour, sexual behaviours, access to clean needles and syringes, access to health care, service needs, and medical service use (e.g., self-reported visits to primary care/outpatient clinics, Emergency Department, detoxification, methadone maintenance, ambulance use and hospital admissions). Participants were reimbursed $20 CDN for each study visit, at which time referrals were provided for medical care, HIV/AIDS care, available drug and alcohol treatment and counselling as needed. The VIDUS study participants may not be representative of all IDUs because those in the lowest socioeconomic group are overrepresented in this study sample. However, it is this group that is most in need of innovative interventions.

In the present study, a total of 250 individuals were recruited in the order in which they appeared for their regularly scheduled appointment for VIDUS. A subsequent appointment for the quality of life study was scheduled and participants were paid $10 CDN in each session of the present study. Data from nine participants were excluded because of missing data or because they were deemed, at the time of data collection, to be too impaired to focus on the research tasks. The final sample consisted of 241 IDUs ranging in age from 19 to 61 years (M = 39.4, SD = 9.5 years). There were more males (63%) than females (37%) and most participants (85%) had completed high school. There were no significant socio-demographic or drug using behaviour differences between the 250 recruited individuals and the other VIDUS participants.

The first 50 participants were invited to return for a second session within 6–8 days to collect test-retest reliability data. All 50 participants returned for the second session as scheduled. In the test-retest group, 58% were male and 42% were female. These participants ranged in age from 22 to 59 years (M = 41.7, SD = 9.2). Most of the participants (90%) had completed high school.

The present version of the IDUQOL consists of 21 life domains and builds on the original version first published by Brogly et al. [13]. Many of these domains (e.g., Being Useful, Drugs, Drug Treatment, Harm Reduction and Neighbourhood Safety) are particularly relevant to the physical, social, psychological, occupational, and geographical reality of IDUs' lives. Life domains are each represented on a 5 by 5 inch card, with the name of the domain and a simple representative picture on the front of the card and a description of the domain on the back of the card. Graphic representations were used so that this measure would be more accessible to respondents who do not speak English as a first language or have low literacy skills.

Although the administration of the IDUQOL permitted respondents to subjectively weight the importance of the life domains to his/her quality of life, a review of the literature on importance ratings and weighting [18] as well as an empirical comparison of the utility of weighted versus unweighted scores with the IDUQOL showed that weighting does not improve upon the use of simpler unweighted scores [19]. Thus, unweighted scores are used in the present study, wherein the respondent simply assigned a satisfaction rating for each life domain using a 7-point Likert-type scale ranging from 1 (very dissatisfied) to 7 (very satisfied) and illustrated with seven stylised frowning and smiling faces. Again, visual representation was included as a guide for respondents with limited English or literacy skills. Domain scores were summed and averaged to obtain an overall quality of life score ranging from 1 (very dissatisfied) to 7 (very satisfied).

The SWLS is a 5-item global measure of life satisfaction [20]. Scores range from 5 to 35, with higher scores representing greater life satisfaction. This measure was selected because life satisfaction was seen as a related construct to quality of life.

The RSES is a 10-item measure of global self-esteem [21]. Total scores range from 10 to 40, with higher scores representing greater self-esteem. This measure was selected because self-esteem was seen as a related construct to quality of life.

The MC X2 [22] is a 10-item short form version of the Marlowe-Crowne Social Desirability Scale (MC SDS) [23]. Strahan and Gerbasi reported that it correlates .80 or higher with the MC SDS. The MC X2 provides an estimate of socially desirable responding as a potential source of measurement error. Total scores range from 0 to 10, with higher scores representing higher social desirability in responding. The MC X2 was selected because measures of pervasive characteristics such as social desirability are strongly recommended to assess discriminant validity [24, 25].

In examining criterion-related validity, the following demographic variables were used to create groups expected to differ in their quality of life: stability of housing, sex trade involvement, high-risk injection behaviours (i.e., lending or borrowing needles, daily use of heroin, cocaine, speed, or crack), involvement in a methadone maintenance program or drug treatment program, reporting hospitalization and emergency department attendance or overdose within the previous six months. Predictive criterion variables included: hospitalization and emergency treatment over the six-month period post-data collection. All variables were measured and coded dichotomously.

Ethics approval for this study was obtained from the University of British Columbia and Providence Health Care Research Ethics Boards. Participants met one-on-one with one of three trained VIDUS staff members for a single session lasting approximately 25–30 minutes. Participants were identified only by their VIDUS study ID code on all research forms utilized for this project. All participants provided informed consent and then completed the study measures in the same order (IDUQOL, MC X2, SWLS, RSES). Demographic and predictive criterion data were obtained from VIDUS using the participants' VIDUS ID codes, a use that was disclosed to participants as part of their informed consent. Retest sessions for the sub-sample of 50 participants followed the same consent process, format, and tasks as the initial session.

To be able to use a summed total score on a measure such as the IDUQOL, it is important to demonstrate that the measure shows either strict or essential unidimensionality [26, 27]. Strict unidimensionality denotes the presence of a single common factor whereas essential unidimensionality indicates the presence of a reasonably dominant common factor along with secondary minor dimensions [28, 29].

An exploratory factor analysis using principal axis factoring in SPSS 12.0 was conducted on the 21 items of the IDUQOL to determine whether essential unidimensionality was present and supported the use of the IDUQOL total score. According to Gorsuch's guideline of 5 to 10 cases per item [30], the sample size for the present study (n = 241) was considered adequate for factor analysis of the 21-item IDUQOL. The data met the criteria for Bartlett's Test of Sphericity, χ²(210) = 1488.02, p < 0.001 and the Kaiser-Meyer-Olkin criteria for sampling adequacy, KMO = .88 [31]. The first factor had an eigenvalue of 6.40 and explained 30.5% of the variance in participants' responses. The ratio of the first to the second eigenvalue was 4.3 which exceeds the strict criterion of a ratio greater than 4.0 for evidence of unidimensionality [30, 32, 33]. These results, in addition to a visual examination of the scree plot (see Figure 1) indicated an essentially unidimensional factor structure for the IDUQOL, which supported the use of a total score [32‐35]. Factor loadings ranged from .31 to .71 for all IDUQOL items on a single factor.

Table 1 displays the inter-item correlation matrix for the IDUQOL. The mean inter-item correlation was .26, which Clark and Watson [36] describe as acceptable. Table 2 shows the mean performance and internal consistency of scores obtained by the sample on the IDUQOL, SWLS, RSES, and MC X2. Given the focus in the present study on the IDUQOL, gender differences on scores from this measure were also examined. No statistically significant differences in performance on the IDUQOL were found between men (M = 4.25, SD = 0.96) and women (M = 4.10, SD = 1.02), t (239) = 1.14, p = .20, and the effect size (d = 0.15) is considered small according to Cohen [37].

In addition to an internal consistency reliability estimate, the one-week test-retest reliability estimate for the IDUQOL scores was also computed. Based on the sub-sample of 50 participants who completed the measure twice, the test-retest reliability estimate was .78, with correlations for each domain across the two sessions ranging from .32 to .67. Table 3 shows the test-retest correlations for all domains.

Table 4 shows the correlations of the IDUQOL total scores with the dichotomously scored criterion variables. Of the statistically significant correlations, all were in the expected direction. That is, lower IDUQOL scores were related to unstable housing, sex trade involvement, borrowing and lending needles, daily use of heroin and speed, and overdose in the past six months. The IDUQOL scores did not correlate significantly with daily use of cocaine or crack, methadone or drug treatment, emergency treatment, or hospitalization within the six months prior to, or following, the initial test session.

IDUQOL scores are based on a wide range of domains that encompass social, physical and emotional realms, and therefore, as a total score, might not correlate significantly with specific criterion variables. To explore this possibility, analyses were carried out at the domain level, matching available criterion variables with relevant IDUQOL domains. For example, the criterion variables of engaged in sex trade and Rosenberg Self-Esteem Scale scores were correlated with the Feeling Good about Yourself IDUQOL domain score. Table 5 shows the correlations of selected IDUQOL domain scores and corresponding criterion variables.

Table 6 shows the correlations of IDUQOL total scores with the SWLS, RSES, and MC X2. The convergent measures (SWLS, RSES) showed moderately high correlations with the IDUQOL as would be expected between constructs that are related but not the same. The correlation between the IDUQOL and the discriminant measure (MC X2) was in the low to moderate range and thus acceptable [38]. As expected, the convergent measures were both more highly correlated with the IDUQOL total score than was the discriminant measure. Correlations were also conducted between the MC X2 and both the SWLS (r = .35) and RSES (r = .41). Because the relationship between the IDUQOL and the convergent measures could be due to the common influence of social desirability bias, partial correlations between the IDUQOL total scores and the SWLS and RSES, controlling for MC X2 scores, were conducted. These are reported in Table 6.