Academic journal article Journal of Drug Issues

Assessing Intervention Efficacy: An Example Based on Change Profiles of Unprotected Sex among Drug Users

Academic journal article Journal of Drug Issues

Assessing Intervention Efficacy: An Example Based on Change Profiles of Unprotected Sex among Drug Users

Article excerpt

Over 700 active drug users recruited in East Harlem, New York City, to participate in an AIDS prevention project were interviewed on two occasions, 6 months apart, to assess changes in HIV-related risk behaviors. This paper presents an example of a method for analyzing patterns of risk behavior change over time as a means of comparing the effectiveness of two interventions. Results described in this paper focus on the number of unprotected sex acts reported in the 30 days prior to each interview and reflect five distinct patterns of risk level over time (i.e., a decrease, an increase, remaining at low risk, remaining at high risk, or no sexual activity at either time). Bivariate and multivariate analyses indicated that (1) compared to persons at high levels of unprotected sex at follow-up (time 2), those who remained at a low level or decreased were more likely to be HIV positive; and (2) age, living alone, and having a stable source of income were also significant predictors of risk pattern. Risk pattern was not associated with type of risk reduction intervention (standard or enhanced) or with drug treatment (yes or no) between baseline and followup. Implications of the findings were discussed with respect to (a) the assessment of efficacy of AIDS prevention interventions and (b) the analysis of risk behavior changes over time.

Introduction

Most longitudinal studies of HIV risk behavior change examine change as a group phenomenon. That is, change is generally reported as a group mean or as a percentage for dichotomous variables. Although the assessment of group change is valuable in evaluating the overall effect of an intervention, it may obscure changes in subgroups of interest. Typically, such subgroups are based on gender, ethnicity, age, or other demographic and baseline characteristics. It is relatively rare that subgroup analyses are based on risk behavior change categories. One reason for this is that if change is the outcome of interest, a measure of change in the same behavior or outcome cannot be used as an independent variable. However, some analytical techniques that are designed primarily to characterize population subgroups, rather than predicting change, do use change scores as one subject characteristic. Cluster analysis and discriminant function analysis have been used in this manner (Rapkin and Luke 1993). In this paper, we describe the use of such a technique to evaluate the efficacy of an HIV prevention intervention using levels of change to characterize a sample of drug users who exhibit different degrees and patterns of change in unprotected sex over a 6-month period.

Longitudinal research on injecting drug users who participate in intervention programs, are exposed to community outreach efforts, or take part in HIV testing and counseling, supports the conclusion that many of those who are interviewed a second time report reductions in HIV risk behaviors (Booth and Watters 1994). There is evidence that both self-initiated efforts among at-risk populations unrelated to specific risk reduction programs as well as program effects contribute to this overall behavior change (Colon et al. 1995). Furthermore, research indicates that, in general, study participants reduce their drug-related risk behaviors more than their sexual risk behaviors (Colon et al. 1995; van den Hoek et al. 1992). If sexual risk reduction has occurred, changes in the number of sexual partners is more likely than initiation or maintenance of condom use (Calsyn et al. 1992; Chitwood and Comerford 1990; Catania et al. 1989).

It appears that it is more difficult to help people to initiate and/or maintain changes in sexual risk behavior than drug risk behavior. A possible reason for this is that drug use behavior, at least in the settings studied, is often "semipublic" and is, therefore, subject to scrutiny by others and consequently to changes in community norms (Catania et al. 1994; Deren et al. 1993; Friedman et al. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.