Background
The San Diego-Mexico border is one of the most active drug smuggling corridors in the world. San Diego has been designated as a High-Intensity Drug Trafficking Area due to the large quantities of cocaine, heroin, and methamphetamine that are transported to the County from Mexico. Nearly two-thirds of women and more than half of men arrested and booked into jail for crimes in San Diego County in 2010 tested positive for illicit drugs, such as marijuana, methamphetamine, cocaine, and heroin [
1]. It is estimated that the total economic cost of alcohol and drug abuse in the region is more than $240 billion annually, with about $97 billion due to drug abuse [
2]. Patients who visit hospital emergency departments (EDs) may be at particularly high risk for a variety of behavioral risk factors such as illicit substance use [
3].
Screening, Brief Intervention, and Referral to Treatment (SBIRT) is a comprehensive, integrated public health approach for providing a spectrum of early detection and intervention services for substance use in general medical care settings, including the ED [
4,
5]. These settings offer a potential “teachable moment” because patients may have perceptions of vulnerability about their health, regardless of the reason for the visit, and therefore may be particularly receptive to screening and counseling [
6]. Unlike primary prevention that targets non-risk or low-risk users, or treatment services for people already dependent, SBIRT provides early intervention services targeted at individuals who misuse alcohol or illicit drugs, but who may not be dependent. Although individual program frameworks vary, all SBIRT programs share two key components: screening and intervention. Individuals who screen positive for alcohol or drug problems are provided with an appropriate educational or therapeutic service. Most of those screening positive are categorized as relatively low risk and receive a brief intervention, consisting of a time-limited motivational interview done on site that focuses on increasing patient awareness of the risks of substance abuse, feedback on normative use and safe limits, and eliciting motivation to change [
4]. Individuals at moderate- to severe risk are provided brief intervention plus brief treatment (e.g., six face-to-face counseling sessions) or referral to specialty treatment for more intensive support [
4].
Although the SBIRT approach has shown promise for alcohol use [
7‐
9], relatively little is known about its effectiveness for illicit drug use specifically [
10]. An international study reported that brief intervention in primary health-care settings was associated with reductions in self-reported illicit substance use in several countries, with the exception of the United States [
11]. Madras and colleagues [
12] found a 68 percent reduction in self-reported illicit drug use among those exposed to screening and brief intervention services, although their study did not include a control group. A randomized study of opioid and cocaine users screened by peer interventionists during an urgent care visit reported a salutary effect of screening and brief intervention on drug use [
13]. With the exception of the Bernstein et al. study [
13], methodological issues, such as lack of biological confirmation of drug use, short follow-up periods, lack of control groups, and the inability to rule out reactivity to measurement, limit conclusions about intervention effectiveness.
SBIRT is quickly becoming a recommended best practice in a variety of settings, especially in EDs and trauma centers, and billing for SBIRT services is becoming easier as more states activate billing codes. However, rigorous research is needed before SBIRT for drug use is ready for broad universal dissemination [
14]. The present study is one of the first rigorous studies to evaluate the effectiveness of SBIRT for illegal drug use. This randomized controlled trial assessed the effectiveness of the SBIRT approach for outcomes related to drug use among patients visiting EDs at two large, urban, acute-care hospitals in Southern California.
Results
Participation in hair sampling at follow-up
Thirty-one participants at follow-up reported being abstinent from all drugs during the past 30 days. MTs collected hair samples for verification from 14 of the 31 participants (8 Life Shift and 6 Shift Gears participants). Of the 14 that provided hair samples, 10 were found to be abstinent and four were non-abstinent. Hair samples were not collected for the remaining 17 participants for the following reasons: a) nine declined to provide a hair sample at the time of follow-up, b) three were not available because they had moved out of the state and conducted the interview with the MT by telephone, c) three had insufficient body hair, and d) two failed to have hair samples taken for unknown reasons. Missing collection of hair among those reporting abstinence did not differ by condition. Conservatively, all 17 were assumed to be non-abstinent in the analysis of past 30-day abstinence.
Effects of intervention
As shown in Table
2, self-reported past 30-day abstinence from all drug use assessed during the follow-up visit was 12.5 percent for
Life Shift and 12 percent for
Shift Gears, a non-significant difference (p = .888). When results of hair analyses were applied, abstinence rates were 7 percent for
Life Shift and 2 percent for
Shift Gears; the difference between groups was not significant (p = .074).
Table 2
Outcomes of
Life Shift/Shift Gears
using complete cases (n=292) and imputed cases (n = 694)
Past 30 day drug abstinence at follow-up (%)b |
Self-reported | | 12.5 (3.2) | | 12.0 (3.1) | .888 |
Biologically validatedc | | 7.0 (2.5) | | 2.0 (1.2) | .074 |
Self-reported drug use – ASI Composite Score (0-1) |
Complete cases | .059 (.008) | .068 (.010) | .055 (.007) | .095 (.010) | .035a |
Imputed | .070 (.005) | .075 (.006) | .068 (.006) | .085 (.007) | .124a |
Medical problems – ASI Composite Score (0-1) |
Complete cases | .639 (.021) | .176 (.038) | .696 (.020) | .280 (.036) | .404a |
Imputed | .65 (.013) | .219 (.023) | .669 (.014) | .248 (.024) | .627a |
Psychiatric problems – ASI Composite Score (0-1) |
Complete cases | .287 (.031) | .250 (.029) | .292 (0.27) | .243 (.025) | .734a |
Imputed | .264 (.017) | .239 (.016) | .272 (.017) | .228 (.017) | .404a |
Alcohol use – ASI Composite Score (0-1) |
Complete cases | .127 (.018) | .124 (.016) | .106 (.017) | .106 (.015) | .888 |
Imputed | .126 (.011) | .115 (.009) | .126 (.012) | .112 (.010) | .808 |
Past 6 mo. health care utilization |
No. of ED visitsd | .528 (.106) | .828 (.129) | .549 (.111) | .806 (.135) | .826a |
No. of hospitaliz.e | .196 (.094) | .261 (.111) | .146 (.099) | .317 (.117) | .479 |
No. of days hosp.e | 1.26 (.607) | 1.60 (643) | 1.43 (.643) | 2.56 (.681) | .335a |
Driving and traffic risk scores |
Complete cases | .975 (.068) | .831 (.070) | .840 (.064) | .907 (.066) | .057 |
Imputed | 1.00 (.045) | .935 (.042) | .837 (.048) | .897 (.045) | .165 |
Table
2 presents mean ASI drug use composite scores by condition and results of statistical tests of changes. Of interest are the relatively low baseline drug use composite scores in both groups (approximately .06 on a 0 to 1 scale). Among complete cases, the
Life Shift intervention group showed relatively small increases in ASI composite scores for self-reported use of any drugs compared to the
Shift Gears attention-placebo control group. The differential change resulted in a significant group-by-time interaction (p = .035) in favor of
Life Shift intervention effectiveness, in addition to a time main effect. An additional analysis indicated that the differential group change in drug use scores did not differ for marijuana-only users versus users of other drugs (data not shown). When imputed data were analyzed, however, the interaction was no longer statistically significant.
Psychiatric problems, and particularly medical problems, declined over time in both groups at about the same rate, resulting in a statistically significant time main effect that was found with complete cases and imputed data. Alcohol use showed no time main effect or interaction. Considering health-care utilization outcomes, there were no group-by-time interactions; however, both groups increased over time (time main effect) in the number of ED visits and the number of hospital days. Number of hospitalizations also increased in both groups, but did not approach statistical significance.
Results of driving and traffic risk scores showed a marginally significant interaction effect (p = .057), with the Life Shift intervention participants showing greater improvement than Shift Gears participants. This finding was unexpected insofar as those in the Shift Gears condition received the intervention in reducing driving and traffic risks, whereas the Life Shift group did not receive that intervention.
Discussion and conclusions
This study found no support for the effectiveness of the SBIRT approach for illicit drug use. The primary outcome variable, past 30-day drug abstinence, was not significant. Analyses of ASI drug use composite scores using imputed data were also not significant.
Comparing our results to those of others is difficult given the lack of comparable study designs, differences in the types of drug users targeted, and other important methodological differences. Bernstein and colleagues’ randomized trial of brief motivational intervention in clinics for opioid and cocaine users is the most similar to the present study in terms of design [
13]. Those authors reported a 4.6 percentage-point difference in biologically validated past 30-day abstinence rates between intervention and control groups at 6-month follow-up [
13], similar to the 5 percentage-point difference in abstinence rates reported here. They also reported beneficial effects of the brief intervention on ASI drug and medical composite scores. Their results are in stark contrast to ours, insofar as we did not see reductions in ASI drug scores in the SBIRT intervention group. Differences in enrollment criteria, the racial/ethnic composition of participants, the content/intensity of what the control group received, and the type of drug users enrolled make formal comparison between the two study results difficult. It is also noteworthy that the Bernstein study enrollees had much higher ASI drug use scores at baseline (.25 versus .06), and lower ASI medical scores (.56 versus .67) than did our participants. Perhaps the benefits of the SBIRT approach are more greatly realized among those at higher addiction levels.
Our study sample differed from those reported elsewhere in terms of their ASI scores. For example, the drug use composite score in our sample of .056 is lower than the score of .09 reported for a nationally representative sample of those in outpatient treatment programs [
32], and far lower than the score of .25 reported for opioid/cocaine users [
13]. These differences are not surprising given that the present study sample was not in treatment; so one would expect them to have lower composite scores. The current study had a large proportion of marijuana-only users; to the degree that they were less likely to perceive their drug use as a problem due to changing societal norms, their drug use composite scores would be relatively low. Psychiatric problem scores in the current sample, however, were almost twice as high as those among patients undergoing outpatient drug treatment [
32], and medical problem scores were 24 percent higher than among opiate and cocaine users [
13]. These differences underscore the heterogeneity of drug users in terms of their co-morbid mental health and medical status, and underscore the importance of addressing mental health and/or medical needs within a population needing substance use treatment.
There was no evidence that the SBIRT drug intervention had an effect on medical/ psychiatric problems, alcohol use, or health-care utilization. Although results from other studies are mixed, our findings are in line with a meta-analysis that found no statistically significant effect of SBIRT interventions on health-care utilization [
39]. Furthermore, the U.S. Preventive Services Task Force [
40] reported that evidence is insufficient to demonstrate that psychosocial intervention reliably improves non-drug use outcomes for largely asymptomatic patients whose illicit drug use is detected through screening. Time main effects for several of these outcomes were observed in the present study, indicating similar changes in both intervention and control groups. For example, there were decreases in self-reported medical and psychiatric problems often associated with drug use, and increases in health-care utilization for both groups. Although this pattern might seem paradoxical, differences in the measures may partly explain the finding. ASI medical and psychiatric problem scores targeted problems in the past 30 days, whereas health-care utilization measures covered a longer period of time (past 6 months). Furthermore, participants’ medical and psychiatric problems at baseline were likely relatively high, insofar as they were currently patients in an ED visiting for a medical problem. At follow-up, however, there were no patients awaiting care. Researchers suggest that patients may no longer have been troubled by their previous medical problems, and therefore, their problem scores were lower.
Despite cohort maintenance activities, a high dropout rate was a limitation of the study. In the present study, populations more likely to be lost to follow-up included younger people, individuals with more severe baseline drug use, and those with lower baseline drug use avoidance self-efficacy. High dropout is a reported problem in many drug use studies, although other studies similar to the present one have achieved high follow-up rates [
11,
13]. Low participation in hair collection for confirming drug abstinence was also a limitation. Another possible limitation is related to using hair samples to confirm self-reported drug use at follow-up, but not at baseline. Participants were expected to have accurate disclosure at admission into the study [
16,
41], although others have reported that individuals may report drug use at baseline that is not confirmed [
13]. In addition, because the HE delivered the type and level of intervention, he/she was aware of the participant’s assigned condition, so some bias may have been introduced. However, measurement staff members were blind to the participants’ conditions. Finally, a small sample size (n = 97) for some analyses was also a problem.
The methodology of the study has many strengths, including its minimal exclusion criteria. The decision to include participants who were using an assortment of substances and multiple substances was made for both scientific and clinical reasons. Recruitment of varied and polydrug users in research trials has been advocated by experts in the field of substance use disorder research as a “real world” test of an approach, and as a means to bridge the gap between research and practice [
42]. The use of an attention-placebo control group is also a strength, as it allowed us to test the rival hypothesis that improvement in drug use occurred because of the participant’s expectations or the attention received, rather than from the SBIRT itself [
43]. Unexpectedly, the SBIRT group tended to improve more than the placebo group on the measure related to driving and traffic safety. The SBIRT drug use intervention may have been more salient or interesting than the attention-placebo control intervention, bringing about changes in a variety of health and safety areas among those participants who received it. Attempted biochemical validation of reports of abstinence, a multiethnic sample, and rigorous modeling of missing data are additional strengths.
The null results of the present study are disappointing, yet it is premature to conclude that SBIRT cannot work for drug use. Alternative explanations, such as those related to intervention implementation and measurement may have obfuscated SBIRT’s effects. Future studies are needed to rule out alternative explanations and add to the knowledge about SBIRT effectiveness for drug use.
Acknowledgements
We gratefully acknowledge the contribution of Elizabeth Clapp, Jessica Lawrenz, Militza Bonet, Wanda Claproth, Fabian Martinez, Patricia Galindo, David Nguyen, and Dr. Gideon Koren. Grant Support: This study was funded by NIH/NIDA under the American Recovery and Reinvestment Act (ARRA) of 2009, grant #1RC1DA028031-01.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
SW and JC contributed to the conceptualization of the manuscript. SW, KE, and CM participated in data analyses. KE, MH, AS, CBS, EC, TC, and MS participated in the implementation and intervention aspects of the study. JG conducted the hair analyses. All authors read and approved the final manuscript.