Outcome studies indicate that intensive interventions are best suited to high-risk offenders who have relatively more severe criminal dispositions and drug-use histories, but may be ineffective or contraindicated for low-risk offenders (e.g., Gendreau et al., 2001).

This is known as the "Risk Principle" in the criminal justice literature, and is attributed to the idea that low-risk offenders are less likely to be on a fixed antisocial trajectory and are more likely to adjust course readily after a run-in with the law. Therefore, intensive treatment and monitoring may offer little incremental benefit for these individuals, while the cost is substantial. High-risk offenders, on the other hand, are more likely to require intensive structure and monitoring to alter their entrenched negative behavioral patterns.

The greatest risk factors reported in the literature for failure in offender rehabilitation programs are

• a younger age during treatment (typically under age 25),
• an earlier age of involvement in crime (especially violent crime prior to age 16),
• an earlier age of beginning drug use (typically prior to age 14),
• a comorbid diagnosis of antisocial personality disorder (APD) or psychopathy,
• previous failed efforts in drug treatment or a criminal diversion program,
• and first-degree relatives with drug abuse problems or criminal histories (e.g., Gendreau, 1996).

These risk factors are labeled "static" because they are historical in nature and are generally unaffected by clinical interventions. "Dynamic" risk factors, which can be targeted for change during treatment, include such things as antisocial attitudes, criminal associations, and gang membership.

The research program at the Treatment Research Institute (TRI) at the University of Pennsylvania has validated the Risk Principle among drug court clients. With funding from NIDA and the Center for Substance Abuse Treatment, TRI randomly assigned misdemeanor drug court clients either to an intensive level of judicial supervision involving biweekly status hearings in court, or to a low level of supervision in which they were monitored by treatment personnel and only had status hearings as needed in response to sustained noncompliance or serious infractions. The results revealed no differences for participants as a whole in counseling attendance, urinalysis results, graduation rates, or self-reported substance use or criminal activity during treatment or at 6 months or 12 months postadmission (Marlowe et al., 2002; Marlowe et al., 2003).

Importantly, however, the study showed a significant interaction effect, depending on participants' risk status. Participants who met DSM-IV diagnostic criteria for APD or had prior experiences in drug abuse treatment attained significantly greater drug abstinence and were significantly more likely to succeed in graduating from the drug court program when they were assigned to biweekly hearings. Conversely, clients without APD or a prior history in drug treatment performed better when they were assigned to as-needed hearings (Festinger et al., 2002). These same findings were replicated in two additional jurisdictions, in rural and urban communities and serving both misdemeanor and felony offenders (Marlowe et al., 2003; Marlowe et al., in press).

In the replication studies, the magnitudes of the interaction effects were quite large. For instance, misdemeanor participants with a prior drug treatment history provided substantially more drug-free urine specimens during the first 3 months of drug court (11.50 versus 2.67) and were substantially more likely to graduate successfully from the program (83 percent versus 17 percent) when they were assigned to biweekly status hearings as opposed to as-needed hearings. Similarly, felony participants with APD reported engaging in substantially fewer days of alcohol intoxication when they were assigned to biweekly status hearings as opposed to as-needed hearings (0.50 versus 4.83).

The large magnitude of these effects made it necessary to stop the studies prematurely on ethical and practical grounds, and to institute remedial procedures for the high-risk participants assigned to the as-needed condition. The resulting small cell sizes (n = 6 per cell in some analyses) do raise concerns about whether the study samples were adequately representative of drug court clients generally. Because the findings were reproduced in sequential experimental studies and are supported by a previously validated criminal justice theory (i.e., the Risk Principle), one is justified in placing greater confidence in the reliability of the results. Nevertheless, it is essential to replicate this work in new settings with a larger number of participants.

It is also important that the interaction effects, although hypothesized in advance, were not under direct experimental control. TRI is currently conducting a prospective matching study in which drug court clients are randomly assigned to different schedules of judicial status hearings on the basis of an assessment of whether they have APD or a prior drug treatment history. The results of this work will permit an estimate of the effect size and relative costs and benefits of assigning drug offenders to different service tracks in drug court based upon their risk level.

The variables of APD and drug treatment history were found to be the most robust indicators of risk level in these drug court studies. This is quite consistent with prior research on the greatest risk factors for criminal reoffending (e.g., Gendreau, 1996).

It is, however, possible that other risk factors will emerge in future matching studies and permit a more sensitive classification of high-risk and low-risk offenders. Further research is also needed to interpret the influence of prior drug treatment history. It is an open question whether this variable reflects the severity of participants' drug problems, past negative experiences with standard drug treatment, or some other, unknown influence. Further inquiry is needed if we are to gain a definitive grasp of the nature of this interaction effect.