Developmentally inspired drug prevention: middle school outcomes in a school-based randomized prevention trial

Abstract

Prior investigations have linked behavioral competencies in primary school to a reduced risk of later drug involvement. In this randomized prevention trial, we sought to quantify the potential early impact of two developmentally inspired universal preventive interventions on the risk of early-onset alcohol, inhalant, tobacco, and illegal drug use through early adolescence. Participants were recruited as they entered first grade within nine schools of an urban public school system. Approximately, 80% of the sample was followed from first to eighth grades. Two theory-based preventive interventions, (1) a family–school partnership (FSP) intervention and (2) a classroom-centered (CC) intervention, were developed to improve early risk behaviors in primary school. Generalized estimating equations (GEE) multivariate response profile regressions were used to estimate the relative profiles of drug involvement for intervention youths versus controls, i.e. youth in the standard educational setting. Relative to control youths, intervention youths were less likely to use tobacco, with modestly stronger evidence of protection associated with the CC intervention (RR=0.5; P=0.008) as compared to protection associated with the FSP intervention (RR=0.6; P=0.042). Intervention status was not associated with risk of starting alcohol, inhalants, or marijuana use, but assignment to the CC intervention was associated with reduced risk of starting to use other illegal drugs by early adolescence, i.e. heroin, crack, and cocaine powder (RR=0.32, P=0.042). This study adds new evidence on intervention-associated reduced risk of starting illegal drug use. In the context of ‘gateway’ models, the null evidence on marijuana is intriguing and merits attention in future investigations.

Introduction

This report builds from prior work of the Prevention Research Center at Johns Hopkins University Bloomberg School of Public Health, presenting new evidence from a epidemiologically-based randomized prevention trial of two developmental interventions for primary school students. The aim is to estimate a possibly protective intervention influence with respect to risk of early-onset use of tobacco, marijuana, and other drugs (to mean age 13 years).

Prior reports on the evaluation have described the two interventions, which were put into place in 1993, and have presented evidence of their beneficial impact with respect to early target responses. In brief, there were a classroom-centered (CC) intervention and a family–school partnership intervention (FSP), both implemented by the regular classroom teachers in Grade 1. The CC intervention involved an augmentation of the primary school classroom curriculum, with concurrent refinement of the same classroom teacher’s approach to management of unruly and maladaptive classroom behavior. The FSP intervention augmented the usual and customary teacher approach by adding an emphasis on parent–school communication and partnership building. Additional weekly home–school learning, communication activities, and periodic parenting workshops were also included in an effort to demonstrate the benefits of greater parent involvement and collaborative work with teachers to optimize target behaviors and school performance. Early published evidence on the CC and FSP interventions was positive, with intervention-associated improvements in proximal targets of school performance and social adaptation, and with reduced risk of early-onset tobacco smoking to mean age 12 years (Ialongo et al., 1999, Storr et al., 2002).

The participating children have been followed through middle school and the current investigation estimates the hypothesized persistence of an effect on tobacco smoking and a possibly protective intervention effect on a multivariate response profile of other drug use targets. This profile encompasses data on early-onset starting to use alcohol, inhalant drugs, marijuana, and other illegal drugs (e.g. cocaine, heroin), as well as tobacco smoking, with newly gathered data through the middle school years.

The underlying logic, theory, and conceptual models to support early developmental preventive interventions directed toward youthful drug involvement have been evolving for several decades (e.g. see Kellam et al., 1975, Kellam et al., 1983, Kellam et al., 1994a, Kellam et al., 1994b; Cicchetti and Schneider-Rosen, 1984, Kellam and Rebok, 1992, O’Donnell et al., 1995, Ialongo et al., 1999). Despite advancements in developmental research, most school-based drug prevention programs offer drug-specific content to promote resistance skills against peer pressure to use drugs, or to raise levels of awareness about drug hazards and perceived harmfulness of drug use (e.g. Pentz et al., 1989, Prinz et al., 2000). In contrast, this prevention program is inspired by a more general theory of child development in which a youth’s characteristics, conditions, and processes observed at one stage in life might be modified in order to achieve a more health-promoting set of behaviors later in life. For this reason, this type of program is referred to as ‘developmentally-inspired’ prevention. Gottfredson and Wilson (2003) have summarized evidence from several early developmental interventions directed toward youthful drug involvement and have substantiated the potential public health significance of these approaches.

The current investigation has the advantage of additional follow-up assessments completed through the period of early adolescence, when youths enter the period of highest risk for starting to use illegal drugs (e.g. see Wagner and Anthony, 2002). As such, we are able to probe for a general protective intervention impact on all of the above-listed drugs and drug groups under study, and for evidence of specific intervention impact on some but not all drugs. Because we have multiple targets or response variables under study, our approach is multivariate. In specific, we use a multivariate response profile analysis method based on the generalized estimating equations (GEE), which simultaneously estimates intervention effects on all of these interdependent drug responses. That is, within the framework of a single multivariate response regression model we are able to estimate drug-specific intervention impact (e.g. impact on risk of starting to smoke tobacco versus impact on risk of starting to drink alcohol without parental permission). Additionally, we have explored potential effect-modification of intervention impact in relation to baseline characteristics, with specific focus on male–female differences.