Screening, Brief Intervention, and Referral to Treatment (SBIRT): Toward a Public Health Approach to the Management of Substance Abuse

One of the first alcohol screening procedures, the Michigan Alcoholism Screening Test (MAST, 14), consists of 24 yes-no questions that list signs and symptoms of chronic alcoholism. The MAST has been criticized because of its length, its potential for falsification, and its focus on finding cases of alcohol dependence rather than early identification of risk factors. The shorter 12-item MAST (15) and the four-item CAGE screening test (16) increase the feasibility of screening but still maintain a focus on identifying active alcoholics. An added disadvantage of these screening tests is their use of questions measuring “lifetime” symptoms (“have you ever …”), which can produce false positives when the alcohol problems occurred in the past but have since remitted. A disguised screening test based on the patient's history of traumatic injury (17) was developed to deal with the falsification problem, but this was done at the expense of sensitivity and specificity.

A number of alcohol screening tests have been developed for special populations, including women (18, 19) and the elderly (20). The World Health Organization developed the Alcohol Use Disorders Identification Test (AUDIT) (9,2) in order to maximize cultural and linguistic generalizability of screening results. The AUDIT focuses on both hazardous drinking as well as alcohol use disorders. The AUDIT has been well validated across different cultural groups in a variety of countries, and several shorter adaptations have been developed, including the AUDIT-PC, the AUDIT-FAST, and the AUDIT-C, all of which focus mainly on the quantity, frequency and pattern of drinking (21, 22). Finally, several single item screening tests have been developed and validated. Williams and Vinson (23) found that a single question about the last episode of heavy drinking has good sensitivity and specificity in detecting hazardous drinking and alcohol use disorders. O'Brien et al. (24) found that by asking “How many days do you get drunk?” in a typical week, they could identify college students who are at higher risk of alcohol-related injury. To the extent that very short screening tests may motivate clinicians to screen more often, these tests may have value (see, for example, 25). Nevertheless, the value of longer tests is that the patient's responses to questions about drinking and alcohol problems can be the immediate point of departure for a brief intervention, which typically begin with a discussion of specific screening results.

Although not recommended for routine screening, several biological markers have been useful adjuncts to alcohol screening in emergency medicine and criminal justice settings, such as the breath alcohol concentration (BAC), gamma-glutamyltransferase (GGT, a liver enzyme), and carbohydrate-deficient transferrin (CDT). BAC has a short half-life and does not provide information about risk behavior other than to estimate the extent of recent drinking. GGT and CDT have not been found to be sensitive or specific enough to detect heavy episodic drinking (26).

Given the different needs and substance use patterns of adults and adolescents, self-report drug screening tests have generally been designed and validated for one or the other of these populations. Two types of self-report tests have been developed for adults. The first, exemplified by the Drug Abuse Screening Test (DAST; 27), consists of direct obvious or face valid questions about drug use and related problems that yields a quantitative score reflecting the severity of drug abuse. A later version of the DAST reduced the number of items from 28 to 10 without compromising reliability (28). In contrast to screening tests that ask directly about substance use, several tests have been developed to measure risk factors that are associated with actual or potential substance use disorders. However, tests of this type tend be quite long. For example, some of these tests are embedded in the Minnesota Multiphasic Personality Inventory, which contains over 500 items, and thus may not be appropriate to use in health care settings where there is limited time to administer and score the screening test (29).

Recognizing the need for a comprehensive screening and referral test for adolescents, the National Institute on Drug Abuse (NIDA) developed the Problem-Oriented Screening Instrument for Teenagers (POSIT) (30). The POSIT consists of 139 items that generate scores indicating problems in ten functional areas that are related directly or indirectly to substance use disorders: Substance Use/Abuse, Mental Health Status, Physical Health Status, Aggressive Behavior/Delinquency, Social Skills, Family Relations, Educational Status, Vocational Status, Peer Relations and Leisure and Recreation. The test has demonstrated good reliability and validity in adolescents referred to an assessment service for evaluation of substance use problems (31), but is too long to serve as a brief screening test. One option is to use only the Substance Use/Abuse part of the test, which would make it more efficient for screening in general health care settings.

In addition to the multidimensional screening approach used in the POSIT, several shorter instruments have been developed to screen specifically for substance use among adolescents. The Personal Experience Screening Questionnaire (PESQ) (32) focuses primarily on drug use and related problems, but also collects information on other areas of concern, such as psychopathology. Reliability and validity of this 38-item test have been demonstrated in the detection of individuals with different histories of substance use (33). Another adolescent screening test that has been used at adolescent treatment programs is the Substance Abuse Subtle Screening Inventory (SASSI, 34), a 78-item self-report instrument that classifies adolescents as chemically dependent. Although the SASSI was designed to prevent deliberate falsification by using indirect questions, it has not been found to be very accurate (35, 36). Other screening tests have been designed for more specific populations, such as the 42-item Drug and Alcohol Problem (DAP) Quick Screen, which was developed for use by pediatricians (37). Validity data have been reported for the 30-item revision of this test (38). Finally, the CRAFFT (Car, Relax, Alone, Friends, Forget, Trouble) is brief (6 items) and has been validated with adolescents in primary care settings (39). Because of its brevity, it is more likely to be used than the longer instruments described above.

Despite advances in the development of self-report screening tests for specific types of psychoactive substances, there has been considerably less attention to instruments that screen for multiple substances. To address this deficiency, the Alcohol, Smoking and Substance Involvement Screening Test (ASSIST; 11) was developed to screen for at-risk use of psychoactive substances as well as related problems. The test uses a common format to screen for 11 psychoactive substances as well as injection drug use. The scoring procedure estimates the relative importance of these different risk behaviors for the purpose of prioritizing counseling interventions. Although the ASSIST is not able to identify people who exceed risky drinking limits based on quantities of alcohol consumed, these questions can easily be added to obtain country-specific alcohol risk levels.

A major challenge to combined screening for specific substances is provider burden, which refers to the skills and time demanded of the screening agent. A relatively simple procedure that addresses this problem is the CAGE test adapted to include drugs (CAGE-AID). The CAGE-AID was found to be more sensitive but less specific than the CAGE (40). This easy-to-use four-item test nevertheless requires further questioning if the patient scores positive. Thus, efficiency comes at the price of specificity, and screening questions using a lifetime (ever) approach can result in high numbers of false positives.

A variety of biological procedures have been developed to detect recent drug use through urinalysis, hair testing, and saliva tests. These methods are not capable of detecting substance-related problems or even substance use beyond a narrow time window (41). Urine screening tends to be the preferred method because it is less invasive than blood testing and the drugs or their metabolites are present in relatively high concentrations in urine. A recent innovation that facilitates biological screening is the commercial availability of self-contained urine testing kits. These tests are easy to use and provide rapid access to test results, but they can only indicate drug use over the previous few days. Other problems include a risk of false positives by passive drug exposure or ingestion of foodstuffs, and false negatives due to the use of adulterants.

There are a number of important issues associated with screening instruments that should be taken into account in the design of an integrated SBIRT program intended to serve the needs of a defined population. First, the accuracy of most of the self-report tests has been evaluated under research conditions, which tend to maximize the likelihood of good performance. Although most screening tests have been found to be valid, performance is likely to diminish in routine clinical settings. A major concern is the extent to which the results of a self-report test can be deliberately faked or distorted in an attempt by the patient to preserve a respectable self-image in the health care or social service setting. Although self-report measures of substance use tend to be valid and reliable in the aggregate under most circumstances, accuracy in clinical settings depends on the degree of perceived threat in the data gathering situation, the cognitive processes (such as memory) that are required to produce answers to the questions, and the motivation and other personal characteristics of the respondent (42). A second consideration is cost and efficiency. Self-report tests are free or inexpensive, and they can be administered and scored quickly. Nevertheless, medical staff sometimes view even a small addition to their routine as an unnecessary burden. Although some screening tests are relatively brief (e.g., CAGE and AUDIT-C) and can be administered in one or two minutes, others require more time and administration skills. Biological tests can be costly to use on a routine basis and require even more time to administer and score. Nevertheless, they are often seen as being more consistent with routine medical practice, and this may affect their acceptability to both patients and staff. A third issue is cultural sensitivity. Although research has not been extensive, there is no evidence to suggest that the reliability or validity of self-report tests varies across different ethnic groups (9, 11). A final issue is the target group of the screening program. Many of the adult screening tests described in Table 1 have been designed for finding active cases of alcohol or drug dependence, not to identify risk factors for drug or alcohol abuse. These tests (e.g., the DAST) typically avoid direct questions about use of specific drugs, focusing instead on the problems associated with any substance use in the past. Subtle or disguised screening tests (e.g., the SASSI) do not appear to be sufficiently sensitive or specific to identify active cases, but may be useful in screening for risk factors. Comprehensive screening tests like the POSIT and ASSIST are capable of identifying both “caseness” and risk factors, but they take more time to administer and score. Even single item or very brief screening tests like the AUDIT-C require further questioning once the patient screens positive, so screening tests with skip-out instructions like the AUDIT and ASSIST may save considerable time because most patients need not be screened further after negative responses to the first few questions.

In the first systematic review of research on this subject, Bien et al. (43) evaluated 32 controlled studies involving over 6000 patients studied prior to 1992. Brief interventions with problem drinkers were often found to be as effective as more extensive treatments. It was concluded that the course of harmful alcohol use can be effectively altered by relatively brief contacts in contexts such as primary health care settings and employee assistance programs. Kahan et al. (44) reviewed 11 trials of physician-based brief intervention in medical settings and concluded that brief alcohol interventions are effective, and their public health impact is potentially enormous. Twelve randomized controlled trials were reviewed by Wilk et al. (45), who concluded that brief intervention is a low-cost, effective preventive measure for heavy drinkers in outpatient settings. Additional support for these conclusions, based on new analyses of many of the same studies summarized in previous reviews, has been reported by Ballesteros et al. (46). Moyer et al. (47) reviewed studies comparing brief intervention both to untreated control groups and to more extended treatments. They found “further positive evidence” for the effectiveness of brief interventions, especially among patients with less severe problems. In an extensive review of the literature for the U.S. Preventive Services Task Force, Whitlock et al. (13) concluded that behavioral counseling interventions for alcohol misuse among nondependent primary care patients identified by screening are feasible and potentially effective components of an overall public health approach to reducing alcohol misuse.

Most of the studies cited in these reviews were conducted in primary care settings, where the prevalence of alcohol abuse and dependence tends to be lower than what is found in emergency and trauma centers. Emergency departments and trauma centers have been identified as high-yield settings for alcohol screening (12, 48, 49). A large randomized trial of brief interventions in a trauma center (50) found that a brief motivational intervention was associated with decreased alcohol consumption and a reduced risk of trauma recidivism.

In the course of investigating the efficacy of brief interventions with at-risk drinkers, research has also evaluated the extent to which behavior change is related to individual differences among drinkers, the professional training, ethnicity and gender of the intervention provider, and the nature of the intervention itself (13). In general, behavior change is not dependent on provider training or characteristics, but the dependence severity of the drinker does seem to be an important correlate of low response to brief intervention. Regarding the nature of the intervention, skills training, simple advice, and motivational approaches seem to be equally effective. In addition, the interventions seem to be equally effective with adolescents, adults, older adults, and pregnant women.

Despite the general preponderance of positive findings, some studies have shown no differences between intervention and control groups, and many studies report significant reductions in control group drinking that are comparable to those of the intervention group (51). One explanation for this phenomenon is that the screening procedure itself has a motivational effect, although one study found no evidence that assessment alone was responsible for changes in the control group (51). The other explanation is “regression to the mean,” which describes a statistical tendency for extreme values such as heavy drinking to return to the group average over time.

In contrast to the alcohol literature, there have been few studies of brief interventions for drug abuse. Bernstein et al. (52) found that brief intervention in a clinical setting can reduce cocaine and heroin use. Brief intervention appeared to facilitate abstinence at 6 months, even in the absence of meaningful contact with the treatment system. Baker et al. (53) found that both the provision of a self-help booklet and a single session of motivational interviewing were associated with reduced amphetamine consumption among regular users. Two studies (54, 55) have found that general practitioners can reduce excessive benzodiazepine use in their patients using brief interventions such as letters or consultations. Despite these promising findings from controlled studies, and similar positive results from research described below under Brief Treatment, several investigators have reported negative findings from brief interventions with drug users. Marsden et al. (56) evaluated the effect of a stimulant-focused brief motivational interview (relative to the provision of health risk information about stimulants) among adolescent and young adult stimulant users. No significant differences between groups were found for ecstasy, powder cocaine, crack or alcohol. Other research (53) with psychiatric inpatients showed similar lack of effectiveness.

Research on brief interventions for alcohol and other substance users has accumulated rapidly during the past two decades. Not only are the procedures generally effective with a variety of population groups, they can be delivered with equal effect by a variety of health care providers. Less evidence is available regarding the brief interventions for drug users, but several studies show positive effects. An important question that requires further research is the extent to which brief interventions can be made more effective when combined with stepped care strategies that increase the intensity of the intervention for patients who do not respond initially. Among the options are brief treatment and referral to specialized programs serving persons with alcohol or drug dependence.

BT models are consistent with a public health approach in which large numbers of individuals at risk of developing serious alcohol or other drug problems may be identified through primary care screening or through court-mandated treatment (e.g., stemming from DUI arrests). The BT target population has traditionally been those individuals with less severe substance use disorders. However, there is a growing body of evidence to suggest that brief treatments are effective with a wide range of clients, including persons with mild to moderate alcohol dependence and regular marijuana users. BT may also be appropriate for some patients when previous attempts using traditional treatment approaches have failed, when there are insufficient resources (e.g., client time or insurance coverage) available for longer-term therapy, or when there are long waiting lists for specialized treatment. There is no question that BT is more effective than being on a waiting list and could benefit large numbers of clients who are seeking and waiting for longer term care (71).

Four FDA-approved medications are available that physicians can prescribe to dampen craving, reduce heavy drinking, and/or promote abstinence. These are: naltrexone, a μ (mu)-opiate receptor antagonist; depot naltrexone, an extended-release form of naltrexone that is injected monthly; acamprosate calcium delayed-release tablets; and disulfiram (under supervised administration). These medications can be helpful to patients who are struggling to maintain sobriety and for preventing relapse after referral to treatment.

Of particular interest to SBIRT, one recent study looked at whether general internists and primary care physicians could treat alcohol-dependent patients as effectively as addiction specialists, using naltrexone (72). Results indicated that primary care counseling with naltrexone pharmacotherapy is a promising approach that can be effective in selected patients. In addition, the long acting, injectable form of naltrexone that is now available may enhance its use in primary care settings (73).

With the newer medications now available, there is increasing interest in whether alcohol-dependent individuals can be treated successfully with FDA-approved medications by their primary care physicians in routine medical practice. The comprehensive COMBINE clinical trial at 11 sites with nearly 1,400 patients explored a variety of treatment methods alone and in combination-within the context of medical management (74). Alcohol consumption decreased by 80 percent over a 4-month treatment period, which suggests that medical management by primary physicians in routine practice can be of benefit in treating alcohol dependence (75). However, the medical management used in the COMBINE trial was relatively intensive (nine 20-minute sessions), so the minimal level of contact with primary care physicians necessary to manage alcohol-dependent patients has not yet been determined.

The COMBINE Study also found that naltrexone in combination with a brief behavioral therapy delivered by licensed health care professionals is more effective than more intensive behavioral therapy delivered by licensed behavioral health specialists (74).

The Drug Addiction Treatment Act of 2000 established a new paradigm for the medication-assisted treatment of opioid dependence. Qualifying physicians in a medical office or other appropriate settings may now apply to the Substance Abuse and Mental Health Services Administration (SAMHSA) to prescribe and/or dispense opioid medications for treating opioid addiction. Two sublingual formulations of buprenorphine, a long-acting partial agonist of muopioid receptors, have been approved by the FDA for this purpose.

The decision to allow office based treatment of buprenorphine was based on a large body of clinical experience from other countries and the United States (76–78). A Cochrane Review meta-analysis of 13 randomized clinical trials concluded that buprenorphine is an effective intervention for use in the maintenance treatment of heroin dependence (79).

As noted above, progress has been made in the development of a variety of effective screening procedures. Nevertheless, a number of practical. and logistical issues need to be resolved before a screening program can be implemented. These issues relate not only to who does the screening, but when, where, how often and who pays for it. Given the simplicity of most self-report screening tests, they are capable of being administered in variety of different settings and modalities, such as interview, questionnaire, and computer. Dyches et al. (89) describe an interactive telephone technology for screening with primary care patients. Both patients and practitioners had rated the procedure positively. Under some circumstances, impersonal procedures, such as paper and pencil questionnaires and computer-assisted telephone interviews, may be more effective than face-to-face interviews with physicians. Saitz et al. (90) report 50,000 visits a year for screening at an internet web site where positive cases received advice and information, suggesting that an accessible web site can attract high risk drinkers for screening and brief intervention.

In the absence of routine screening with standardized instruments, physicians do not systematically apply NIAAA guidelines regarding hazardous drinking levels (91) and may be selective in whom they screen. One study (92) found that physicians were least likely to initiate discussions about drinking with patients who are white, female and widows. A Danish study of screening by general practitioners (93) reported that physicians did not think all patients should be screened. The major barriers were lack of time and financial incentives, and skepticism that patients wanted to be screened.

Some investigators (94) have argued that primary health care is not an effective or efficient place to conduct alcohol screening. Others have suggested alternative sites for screening and professionals who can conduct it. Anderson et al. (95) identified the clergy as potential providers of screening and brief intervention. Hungerford et al. (96) report that screening in an emergency department can produce high rates of acceptance of counseling about alcohol use. In a rural university emergency department (12), only 3 % of the patients screened or counseled were uncooperative; 70% thought the emergency department was a good place to help patients with alcohol problems. Another potential setting for screening programs is trauma centers. Schermer et al. (97) found that 70% of trauma patients were successfully screened, with less than 1% refusing. Nevertheless, a survey (98) of 50 insurance commissioners indicated 38 states where there are concerns about the possibility that screening will affect insurance payments, which can be denied in many states if the patient has been drinking.

According to Roche and Freeman (99), physician-based secondary prevention efforts based on brief interventions for hazardous drinking have failed at the implementation stage. Barriers to implementation include lack of time, poor diagnostic skills, negative attitudes, and perceptions of role incompatibility (100). In a survey of 711 trauma surgeons (97), 83 % agreed that the trauma center was an appropriate setting but only 25% used formal screening questionnaires and less than one half of problem patients are addressed in their hospital stay. Barriers included cost, time, confidentiality and threats to insurance coverage.

To overcome some of these barriers, other delivery agents (e.g., nurses) have been considered. D'Onofrio and Degutis (101) describe the use of non-physician health promotion advocates (HPAs) to do SBI and referrals in the emergency department.

Another way to expand the use of brief interventions is through internet applications. A review of the small number of web-based interventions (102) found that a demand does seem to exist for this kind of service and the potential impact could be considerable. As noted above, Saitz et al. (90) recorded 50,000 screening visits a year at an internet web site. Positive cases received advice and information, suggesting that an accessible web site can attract high-risk drinkers for brief interventions.

Another barrier to brief intervention is competition for the provider's time once a patient screens positive. Saitz et al. (103) showed that the very act of screening can prompt physicians to increase discussions and provide advice. Brady et al. (104) found that prompting of providers using other means doubled the rate of brief intervention. In a study by Kaner et al. (105), patient and practitioner characteristics predicted who got a brief intervention after screening: males, the unemployed, and technically trained workers were more likely to receive an intervention than females and employed persons. In addition, practitioners with more training and longer practice experience were more likely to deliver interventions. Babor et al. (106, 107) compared two different implementation strategies for Cutting Back, a primary care alcohol screening and brief intervention program for hazardous and harmful drinkers. In one model, medical providers were responsible for delivering interventions. In another model mid-level professionals (usually nurses) acted as the clinic specialists to provide that service. In a sample of 10 health clinics, the mid-level professionals screened a higher percentage of patients than did the medical providers during the best month of program operation (50% vs. 44%) and over all months of operation (24% vs. 19%). Of those patients who screened positive, more patients screened by the midlevel professionals received an intervention than in the provider model (73.1% vs. 57.1%). The ability of clinics to conduct SBI was significantly correlated with both staffing characteristics and organizational factors (e.g., prior experience, organizational stability, number of clinicians trained and the quality of the coordinator's work). Lack of time, staff turnover and competing priorities correlated negatively with SBIRT implementation.

In summary, the primary obstacles to the use of SBIRT in applied settings are: (1) lack of time for overburdened health care workers; (2) training and motivation of professionals to administer screening and intervention to at-risk drinkers; and (3) organizational factors including administrative support and competing priorities. Successful implementation of the technology tends to occur at those sites where clinicians are reimbursed for their services and are well trained for the task. In addition, the extent to which a given delivery model is likely to work best within a managed care organization depends on complex provider and organizational characteristics.

Although there are insufficient data to determine which populations might benefit most from Brief Treatment, a growing literature suggests that BT is effective with a wide range of substance abusing clients. Further, the majority of clients receiving substance abuse treatment stay in therapy for relatively short periods of time (between 6 and 20 one-hour sessions). Although this statistic argues for a greater use of structured BT approaches in current clinical practice, DHHS (71) found that many therapists trained in traditional approaches were resistant to using structured BT models. A related problem is that brief treatment is typically developed, evaluated and delivered in an individual therapy format, whereas traditional treatment tends to be offered in group format because of cost considerations.

The demonstration of several efficacious brief treatment interventions, especially for marijuana dependence, raises questions about how best to engage chronic marijuana users in treatment and how best to maintain improvements following treatment. Unfortunately, very little research has been conducted in these areas. A pilot study was conducted to evaluate a program designed to offer a guided self-assessment (but not treatment) to persons interested in changing their marijuana use. It successfully used a variety of recruitment strategies to attract participants, including posters, radio and newspaper ads, and outreach at various community events (108). The check-up program offered a useful method for reaching non-treatment-seeking heavy marijuana users, and at follow-up program participants reported a significant reduction in the frequency of marijuana use when compared to those who just got information. These results suggest that standalone programs that provide discrete treatment to regular marijuana users may be feasible and can reach large numbers of clients if they are properly designed and advertised.

Training in how to conduct screening and brief interventions is clearly a vital component in assuring effective implementation of SBIRT components. Introducing new screening and prevention activities into primary care practices and other settings presents significant challenges to professional training and continuing education. Medical schools and residency programs devote limited faculty resources and curriculum time to substance abuse (109–111) and many professionals feel inadequately trained when faced with patients who have substance-related problems (112, 113). Barriers to adequate coverage of alcohol and drug-related problems in both medical schools and continuing professional education include traditional attitudes about the moral culpability of chronic alcoholics, confusion as to whether problem drinking is a medical or psychiatric concern, lack of faculty role models, lack of training materials, and role ambiguity regarding who is responsible for screening and intervention (114, 115). Research on medical education has shown that training can be effective in improving students' and physicians' knowledge and skills in addressing alcohol issues (106, 116–118), but changes in knowledge may be easier to produce than changes in attitudes and behavior (119). A review of the components and outcomes of medical education in substance-related disorders concluded that the selection of a combined didactic and interactive educational strategy may be the most cost-effective learning strategy, but there is little empirical evidence to support this approach (119).

Although some progress has been made in the development of SBIRT for medical practitioners, medical students, and health care organizations (1, 120, 121), a necessary step toward dissemination is the development of successful training packages that include program implementation procedures. Babor et al. (106) found that following a relatively short (3-hour) workshop and subsequent supervision, physicians experienced an increased sense of confidence in performing screening procedures. In addition, non-physician clinicians perceived fewer obstacles to screening patients after receiving the training. When delivered in the context of a comprehensive SBIRT implementation program, training was effective in changing providers' knowledge, attitudes, self-efficacy and practice of screening and brief interventions for at-risk drinking. The results are consistent with other studies of provider behavior (116, 122, 123) which show that health care providers trained to deliver a brief alcohol intervention will counsel their at-risk patients when cued to do so and when supported by a primary care office system. Adams et al. (123) found that a 2.5-hour training doubled the rate of alcohol interventions in high-risk primary care patients. Wilk and Jensen (124) used standardized patients (i.e., actors who play the role of symptomatic patients) to train residents to use SBI. After training more residents conducted screening and brief interventions. Gomel et al. (125) compared three strategies to market and train primary care physicians. Tele-marketing was more cost-effective than academic detailing and direct mail in promoting uptake of an SBI package. Roche et al. (117) compared two educational programs to train medical students; interactive training was no more effective than traditional didactic lectures in developing knowledge and skills. These studies suggest that SBIRT training can be effective in providing skills, increasing self-efficacy, and changing provider behavior.

In summary, training programs have been developed and adapted to specialty settings (e.g., physicians in primary care clinics). Educational materials for use with problem drinkers have also been developed. Manuals, pamphlets, and books have been written to help train professionals in the process of SBI. Research on all of these training packages suggests that they increase knowledge about drug misuse, but they vary in their ability to change provider behavior. More research is needed on how increased knowledge translates into behavior changes and what factors help to sustain those behavior changes.

SBIRT costs will vary, depending on the perspective from which costs are calculated, e.g., the provider's, the payer's, the patient's, or society's perspective. For financial management purposes, the total costs of SBIRT services can be broken down into their components, e.g., screenings, information packets, counseling sessions, and case management. From the provider's perspective the cost of brief interventions depends primarily on the nature and severity of the client's alcohol or drug problems, the number of sessions that comprise the interventions, the personnel delivering brief interventions, the resources to produce and deliver interventions (and treatments) and the settings in which brief interventions are provided. Providers must also consider the one-time costs of developing and starting the service plus any on-going continuing costs such as continuing education of staff. From the client's perspective, the cost of SBIRT includes the amount the client pays for the intervention beyond the premiums for health insurance, as well as time and transportation costs to the site where interventions are furnished. From a payer's perspective, the cost of brief interventions might be defined as the amount paid for the service minus any financial benefit that may accrue from the reduction of future costs resulting from the service. From society's perspective, the cost of brief interventions is expressed in terms of the market value of the best alternative use to which labor, capital, and other resources may be put (i.e., economic or opportunity costs).

Given the various perspectives that could be used, it is not surprising that published estimates of the costs of SBIRT vary considerably. For example, Zarkin et al. (126) estimated screening costs at $0.42 per patient for a 2-minute screen versus $16 per patient by Gentilello et al. (127) and $497 per patient by Kunz et al. (128). Given the fact that fewer than 30 percent of patients screened are referred for brief interventions, efforts to reduce the initial screening costs can significantly reduce the overall cost of providing alcohol SBI. There is also broad variability in the costing methodology used in the literature. For example, brief intervention costs have been reported at $2.59 per patient (126), $135 per brief intervention session (128), and $0.59 median per member per month (insurance premium cost) (129). Obviously, the underlying variability of the SBI programs is a primary cause for the variation in cost estimates, but the lack of a consistent costing methodology contributes to the variability and limits the usefulness of cross-program comparisons.

Furthermore, current SBI cost-effectiveness and cost-benefit research often presents cost results without a detailed description of the costing methods used. Many of these studies do not adequately address how and what was actually costed (e.g., 130). Authors often take national wage averages and estimate the amount of time for services (e.g., 127). The most thorough cost estimate comes from the Cutting Back study (126), which used activity-based costing to separate start-up costs from ongoing implementation costs, a distinction overlooked by previous studies. Cutting Back is the only study to compare costs across providers and is also the first to cost different models of implementation. However, the SBI models studied by Zarkin et al. (126) were implemented exclusively for the Cutting Back research project, and therefore the authors were forced to make judgments as to which costs would likely be retained in a non-research setting.

Cost effectiveness analysis (CEA) enables decision makers to compare the economic merits of alternative types of service, such as brief interventions and standard care, which represents the care that clients would ordinarily receive. Kunz et al. (128) found cost-effectiveness ratios for brief interventions administered in a hospital emergency department of $258 for a one unit reduction in the follow-up AUDIT score, $219 for a decrease of one drink per week, and $61 for a one percentage point decline in the follow-up probability of heavy drinking. In a study that applied estimates from published studies to Australia, Wutzke et al. (130) found that brief physician advice to at-risk drinkers resulted in additional years of life from fewer accidents. Dividing the cost of the intervention by the number of life-years saved yielded a cost of approximately Aus $1,873 per life-year saved. CEA does not, however, provide definitive recommendations on which program should be adopted. Rather, it provides decision makers with evidence on the relative benefits and costs of one program compared to another. For this reason, CEA alone is often not enough to justify adoption of a new program.

Unlike CEA, CBA places a dollar value on all outcomes and directly compares to the dollar value of a program's outcomes to the dollar value of its costs. As a result, CBA often provides definitive answers on which programs should be adopted. The program with the largest dollar benefit after accounting for costs should unambiguously be adopted. There are various methods with which to compare the benefits of a program to its costs, including: net benefit measures in which costs are subtracted from benefits; return on investment in which the benefits are expressed as a percentage return to the investment represented by the program costs; and the benefit cost ratio in which benefits are expressed as a ratio of the costs. The choice of CBA measure is largely determined by the audience, with return on investment often appealing more to business or corporate audiences and net benefit or benefit-cost ratios appealing more to academic audiences.

The CBA evidence on SBIRT is generally very favorable. In a randomized trial of brief interventions administered in physician offices, Fleming and colleagues (131) found that a group receiving a brief intervention not only had significant reductions in alcohol use, they also had fewer hospital days and fewer emergency department visits. The intervention cost $205 per person ($166 from the clinic perspective and $39 from the client's perspective) and saved $712 in health care costs. The benefit cost ratio of 4.3 suggests a $43,000 savings in future health care costs for every $10,000 spent for early intervention. The benefit cost ratio increased to 39:1 after factoring fewer motor vehicle and legal events into the analysis.

In a CBA using published sources, Gentilello et al. (127) estimated that the screening and brief alcohol interventions provided to injured patients treated in an emergency department or admitted to a hospital together cost $54 per patient, or $16 plus $38, respectively. The net cost savings from the screening and intervention was estimated at $89 per patient, or $330 for each patient receiving an intervention (27 percent had a positive screen). The benefit, in the form of reduced direct health costs, resulted in a savings of $3.81 for every $1.00 spent on screening and intervention, for a benefit cost ratio of 3.8:1. If interventions were routinely offered to injured adult patients nationwide, it was estimated that the potential net savings might approach $1.82 billion annually.

In a retrospective study of admissions to the Naval Medical Center in Portsmouth, Virginia, Storer (132) estimated that intervention patients had significantly lower hospital readmission rates than other patients. The lower readmission rate for intervention patients alone generated an estimated savings of $606,400, for a total cost of $31,500 (benefit cost ratio of 19:1), for an average cost of $154 for 205 brief interventions.

Although the findings support the use of certain SBIRT components on economic grounds, the studies should be used cautiously. The cost effectiveness of SBIRT may vary considerably, depending on how the technology is applied. If a program is aimed at a selected, high-risk portion of the population (e.g., emergency room patients with injury or trauma), a higher rate of risky drinkers will be identified than in a “population approach” (e.g., all members of an HMO), where a cross-section of the entire population is screened (133). This issue will affect the rate at which people receive an SBIRT service and the economic efficiency of any such operation. Similarly, the potential for cost savings is much greater among a higher risk portion of the population. The labor cost of personnel designated to screen and conduct brief intervention, and whether SBIRT is their sole function or is incorporated into other functions will affect cost effective calculations. Additionally, the extent of the intervention-whether one five-minute session at the time of the screening, or multiple sessions of longer duration on different days-will significantly affect both treatment costs and costs incurred by patients.