Sensitivity and Specificity

Sensitivity is a screening instrument's capacity to identify true cases of the target condition in a given population. The closer to 100 percent of those with alcohol and other drug problems that a screen identifies as positive for that condition, the more sensitive the test.

Specificity refers to an instrument's ability to identify people who do not have the disorder. False positives (identifying people who do not have the disorder as having it) tend to increase as sensitivity increases, and false negatives (missed cases) tend to increase as specificity increases. Because screening instruments are imperfect, balancing sensitivity against specificity is a situation-specific issue. Generally, for screening in primary care, sensitivity should be emphasized over specificity -- that is, it is more important not to miss true cases than it is to assess further some patients who ultimately turn out not to have a substance use disorder. A positive screen can usually be confirmed or refuted with further history taken on the spot or, if necessary, evaluation by a substance abuse specialist. The screening instruments recommended by the Consensus Panel achieve a reasonable balance between sensitivity and specificity (see Appendix C).

Most screening instruments have been designed for substance abuse treatment populations, not primary care populations. The four-question CAGE questionnaire (Ewing, 1984) and the Alcohol Use Disorders Identification Test (AUDIT) (Babor et al., 1992), however, have been extensively tested in primary care settings, and a number of other studies of outpatient, substance abuse treatment populations support the practice of applying substance abuse screening instruments to primary care populations (Buchsbaum et al., 1991, 1995; Bohn et al., 1995; Barry and Fleming, 1993; Saunders et al., 1993). The CAGE questionnaire is reproduced below, and the AUDIT appears in Appendix C.

Cost

Costs of administering a screen depend on who does the screening (e.g., physician, nurse, nurse practitioner, or physician assistant), how long it takes, and what special training (if any) is required; whether the instrument can be self-administered by the patient via pencil and paper or computer; and how long it takes to score the instrument.

Ease of Administration

The written questionnaire format is self-explanatory; the interview format consists of a clinician's asking the patient a set of predetermined questions. Computerized versions of validated paper questionnaires such as the CAGE are growing in popularity, and preliminary studies on the effectiveness of this approach are promising (Barry and Fleming, 1990). A study of adolescents found that when 15-year-olds were asked about past-week alcohol use, 10 percent responded positively to a computerized questionnaire, but only 5 percent to a paper questionnaire (Paperny et al., 1990). Across populations, however, studies have shown that similar results were obtained regardless of the form of the test (National Institute on Alcohol Abuse and Alcoholism, 1993).

Computers also can reduce the time needed for manual scoring and keep track of who has been screened and when. In addition, some computerized screens like the Diagnostic Interview Schedule format (Blouin et al., 1988) will automatically ask selected assessment questions if the score on screening is positive.

Patient Acceptance

Simply raising the subject of substance abuse with patients can be useful. Evidence indicates that asking questions about alcohol or other drugs "primes" patients to disclose information and results in a two- to threefold increase in their stated intention to discuss substance abuse problems with their health care provider in the future (Skinner et al., 1985).

While opinions vary about whether to integrate substance abuse screening into a standard history, asking potentially sensitive questions about substance abuse in the context of other behavioral and lifestyle questions appears to be less threatening to patients. Studies have found that screening for alcohol-related disorders is more acceptable to patients if it is part of a comprehensive health-risk evaluation that covers topics like exercise, diet, weight control, and medication use (Allen et al., 1995). Placing the questions within the larger context of preventive health care can help both patient and clinician feel more comfortable, reduce any perceived stigma or bias about the questions, and decrease anxiety in the patient.

Members of the Consensus Panel have learned that this finding holds true when screening for use of illicit drugs as well (Fleming and Barry, 1991). Primary care clinicians with experience in substance use screening also report that discussing problematic use can help foster the ongoing relationship between patient and clinician.

Alcohol Screening Instruments

Alcohol screening instruments question patients about how much and how often they drink and/or the consequences of their drinking. Answers to quantity/frequency questions indicate whether a patient was, is, or may be at risk for becoming a problem drinker, a binge drinker, and/or an alcoholic, distinctions important in determining the clinician's response. A hallmark of alcoholism (and drug addiction) is continued use of a substance despite adverse consequences. Questionnaires focusing on consequences generally are quite successful in detecting dependent users; without quantity/frequency questions, however, these instruments tend to miss early stage problem drinkers and at-risk drinkers.

Since no single screening instrument can be used with all primary care patients, clinicians will want to select those options that best meet the needs of their patient population. For patients with low literacy skills, face-to-face interviews where the clinician asks the questions and documents answers will best elicit information. Regardless of the information-gathering technique, however, clinicians are relying on self-reports with no assurance that answers are truthful. At this time, there is no viable alternative to self-reports in the primary care setting (Institute of Medicine, 1990), although urine tests (discussed further below) can often detect recent use of some common illicit drugs, and liver function tests may show liver damage, suggesting excessive alcohol consumption. Since denial is a major symptom of dependence, the validity of self-reports is frequently an issue for those patients with alcohol or drug problems. In this situation, when the clinician suspects that a patient is not responding honestly, she may, with the patient's permission, seek information from such collateral sources as the patient's spouse, parents, and siblings. To assist primary care clinicians with screening instrument decisions, the Consensus Panel recommends the following widely used instruments for the primary care setting.

To screen for alcohol problems using a self-administered written questionnaire, a brief instrument like the AUDIT is appropriate, particularly where the expected reading level and comprehension of written English are not likely to be problematic. The AUDIT takes about 2 minutes to answer (Hays et al., 1993) and about 15 seconds to score. If the screen will be administered by a clinician, the CAGE, supplemented by the first three quantity/frequency questions from the AUDIT, is recommended. This combination will increase sensitivity for detection of both problem drinking and alcohol dependence because it includes questions about both alcohol consumption and its consequences. Self-administering the CAGE alone takes about 30 seconds (Hays et al., 1993).

Drug Screening Instruments

Although screening for drug use in the primary care setting can make patients and clinicians uncomfortable, asking about illicit drug use is as important as asking about other personal practices (such as sexual practices that put patients at higher risk for sexually transmitted diseases) that can affect a patient's health.

Of the drug abuse screening instruments, CAGE-AID (CAGE Adapted to Include Drugs) is the only tool that has been tested with primary care patients (Brown and Rounds, 1995). Like the CAGE, CAGE-AID, reproduced below, focuses on lifetime use. While those patients who are drug dependent may screen positive, adolescents and those who have not yet experienced negative consequences as a result of their drug use may not. For this reason, the Consensus Panel recommends asking patients, "Have you used street drugs more than five times in your life?" In Panelists' experience, a positive answer indicates that drugs may be a problem and suggests the need for in-depth screening and possibly assessment.

Because the questions were originally developed for alcohol, the CAGE-AID will not apply to every illicit drug or drug user. It is, however, a useful starting point. As with the CAGE, the Panel recommends that one positive answer prompt further evaluation.

The Panel recommends that clinicians treating patient populations at high risk for drug abuse ask their screening questions regarding alcohol and drug use in combination. (This high-risk group includes those with psychiatric, behavioral, demographic, familial, social, or genetic risk factors that increase the likelihood of drug abuse. Red flags include work-related, marital and family, or legal problems. See Chapter 1, Figure 1-2.) Patients may view questions about drug use paired with questions about alcohol as less onerous than questions about drug use alone.

Supplementary Laboratory Tests

Although several laboratory tests can detect alcohol and other drugs in urine and blood, these tests measure recent substance use rather than chronic use or dependence. At this time, there is no test like the blood sugar test for diabetes or the blood pressure test for hypertension to identify substance use disorders. For this reason, the Consensus Panel does not recommend the routine use of laboratory tests as screening tools in the primary care setting (Babor et al., 1989; Beresford et al., 1990; Bernadt et al., 1982). Laboratory tests, however, may be useful during the assessment process to confirm a diagnosis, to establish a baseline, and later, to monitor progress (Schuckit and Irwin, 1988). Positive test results can be a powerful incentive for changing behavior or motivating patients to accept referrals for treatment.

For some adolescents, a drug test may be a useful supplement to the screening instrument, especially if changes have occurred in school performance, sleep patterns, weight, mood, or social group. Again, depending on the clinician's expertise and available resources, urine tests can be done in the primary care setting or can be referred out to a drug treatment specialist.

Pregnant Women

It is generally accepted that quantity/frequency criteria should be lower for females than males and that pregnant women should abstain from all alcohol and other drug use. Fetal alcohol syndrome is the most common preventable cause of mental retardation (Abel and Sokol, 1991; Centers for Disease Control and Prevention, 1993). Opiates and cocaine have been implicated in intrauterine growth retardation, premature births, neurobehavioral and neurophysical dysfunction, birth defects, cardiovascular problems in mother and fetus, spontaneous abortion and fetal compromise, vascular disruptions, and increased risk for infectious diseases including human immunodeficiency virus (HIV) (Bandstra and Burkett, 1991).

Because of the potential risk to the fetus, primary care clinicians should ask all pregnant patients about their drug use. The Panel recommends asking directly, "Do you use street drugs?" If the patient answers yes, advise her about possible negative effects on the fetus and recommend abstinence.

Of the alcohol screening instruments that have been modified for pregnant women, the TWEAK (Russell, 1994) (a phonetic acronym for its five questions: "tolerance," "worried," "eye-openers," "amnesia," "cut down") has been found to be the most effective for this population, for whom any use is relevant (Chan et al., 1993). Based on best clinical judgment, the Panel recommends the use of the TWEAK (reproduced below) for pregnant patients in the primary care setting.

Older Adults

A recent study found that for patients age 65 and older, the prevalence of hospitalizations for alcohol-related medical conditions and for myocardial infarctions are similar (Adams et al., 1993). As high as the numbers are now, projections of the future prevalence of alcohol-related problems indicate that the problems among older adults will increase appreciably, especially when the Baby Boom generation turns age 60. To ensure that older adults receive needed intervention services, stepped-up identification efforts by primary care clinicians are essential (DeHart and Hoffmann, 1995). Since warning signs of substance abuse (e.g., sleep problems, falls, and confusion) can be easily confused with or masked by other concurrent illnesses and chronic conditions associated with aging, the Consensus Panel recommends that all adults age 60 and older be screened for alcohol and prescription drug abuse as part of their regular physical examination. At the very least, those older adults undergoing key life transitions (e.g., death of a spouse, retirement, moving, or cessation of caretaker responsibilities) should be screened.

The CAGE and the Michigan Alcoholism Screening Test -- Geriatric Version (MAST-G) (Blow et al., 1992) are alcohol screening instruments that have been validated for use with older adults. The Consensus Panel recommends the use of the CAGE, again with a cutoff score of 1. The lower threshold is particularly important for this population because "age-related physical changes . . . can cause older people to develop more severe intoxication and subsequent problems at lower levels of consumption" (American Psychiatric Association, 1994a, pp. 201-202). There is also "some evidence of increased neural sensitivity to single doses of alcohol with age" (American Medical Association, 1995, p. 5).

Since the MAST-G was developed specifically for older adults, it provides a sound screening option for clinicians willing to spend the time required to administer this 24-item test, reproduced in Appendix C. Although the AUDIT has not been evaluated for use with older adults, it has been validated cross-culturally. Since there are few culturally sensitive screening instruments, the AUDIT may prove useful for identifying alcohol problems among older members of ethnic minority groups.

Individuals with chronic health problems also may be using a large number of prescription drugs, which can cause complications when combined with alcohol and other drugs. To screen for prescription drug use, a clinician can ask questions such as

• "Do you see more than one health care provider regularly? Why? Have you switched doctors recently? Why?"
• "What prescription drugs are you taking? Are you having any problems with them?"
• "Where do you get your prescriptions filled? Do you go to more than one pharmacy?"
• "Do you use any other nonprescription medications? If so, what, why, how much, how often, and how long have you been taking them?"

If the clinician suspects that prescription drug abuse may be occurring and the older patient is confused about her prescriptions, seeing more than one doctor, using more than one pharmacy, or seems reluctant to discuss her use, assessment is warranted.

Health Care Professionals

Health care professionals are not exempt from substance abuse problems and should be screened according to the same protocols applied to the larger primary care population. Limited histories should be obtained from all, and a thorough screening done if the provider is being prescribed a mood-altering drug -- especially when anxiety, depression, and generalized physical complaints are presented. Interventions with this population may be challenging because health care professionals may be convinced that they know about substance use, which they think somehow makes them immune to this problem (Sullivan et al., 1988). While the incentive to complete treatment is compelling -- a license and professional reputation are in jeopardy -- the high stakes may also make it unlikely that they will admit to alcohol or drug abuse on a simple screening. Providers also should watch for physical or psychological signs of substance abuse or behaviors like excessive prescribing or personal use among their colleagues.

Adolescents and Young Adults

Because epidemiological evidence indicates high risk among adolescents and young adults and since early intervention among this group can greatly reduce future health and other social costs, primary care clinicians should routinely screen these patients. According to the American Medical Association's Guidelines for Adolescent Preventive Services (GAPS), all adolescents should be asked annually about their use of alcohol, tobacco, and illicit drugs and about their use of over-the-counter and prescription drugs for nonmedical purposes, including anabolic steroids (Elster and Kuznets, 1994). However, since many teens do not receive annual physical examinations, the Panel recommends that screening occur every time they seek medical services, including visits necessitated by acute illness and accidents or other injuries.

Although the routine use of urine toxicology as part of the screening process of adolescents is not recommended, there are important exceptions. When there is a clinical reason to suspect a substance abuse problem (e.g., recent onset of an emotional or behavioral disorder, a change in school performance, or unexplained need for large sums of money), urine tests can be a prudent adjunct to the screening questions. Adolescents should not be tested without their knowledge and consent, except in a medical emergency. The knowledge that a test will be conducted sometimes prompts more honest replies, although this is not always the case.

If any of the following risk factors or "red flags" are revealed during questioning and examination, the adolescent should be referred to a substance abuse treatment specialist with expertise in adolescent issues for a comprehensive assessment.

Asking the Questions

The Consensus Panel believes that both physicians and nonphysicians can reliably screen for alcohol problems. Expanding the pool of people who screen to include nurse practitioners and physician assistants increases the likelihood that patients who should be screened are. Regardless of their professional positions, the clinicians should have proven screening skills: Early screening by unqualified people can lead to false reporting, which becomes part of the patient's record. Those screening should be familiar with the questionnaire and its interpretation, demonstrate considerable interviewing skills, be able to establish rapport with the primary care patient population, and be sensitive to the potentially stigmatizing nature of screening for alcohol and drug problems.

How the questions are asked tends to be more important than who is asking. One study demonstrated, for example, that the sensitivity of the CAGE questionnaire is dramatically enhanced by an open-ended introduction: "Please tell me about your drinking" (Steinweg and Worth, 1993). Some problem drinkers and illegal drug users may feel embarrassed and guilty about their use; others may respond with hostility to questions raising the possibility of an alcohol or drug problem. To overcome discomfort with alcohol and drug screening questions and increase the likelihood of honest answers, clinicians should pose screening questions and accept patient responses matter-of-factly without judgment. Some clinicians report that assumptive questioning yields more accurate responses: "When was the last time you were high?" for example, is a better question than "Do you drink?" Other helpful questions are, "At what age did you first use?", "At what age did you use most frequently?", and "How many times did you use last month?" Ensuring privacy during the screening also reassures patients that the information they provide will be kept confidential and enhances the rapport between patients and clinicians.

Since screening also can reveal that a member of the patient's family has problems with alcohol or other drug use, clinicians should be sensitive to this possibility. The ongoing, long-standing contact with patients and their families that many primary care clinicians enjoy presents a unique opportunity to support non-using family members who are upset by a spouse's, child's, parent's, or sibling's substance abuse problem, confused about how to proceed, and exhausted from covering up or attending to the problem on their own. These relationships also smooth the way for clinicians to discuss possible substance abuse among other family members and devise a plan for intervening with all those who may be involved. In discussions like these, it is important to assure the patient that confidentiality will be maintained (see Appendix B).

Effective implementation of a screening system will require ongoing training, monitoring, training supervision, and attention to issues of reliability, empathy, appropriate responsiveness, and consistency over time. Use of a well-validated screening questionnaire reduces the risk of personal bias in interpretation.

Documenting Screening

It is important to remember that a positive screen does not constitute a diagnosis, even if the screen suggests a high probability of risky alcohol- or drug-related behavior. If and when the positive screen is confirmed by further assessment and discussed with the patient, clinicians should then explain the implications of including positive screening results in the medical record. While medical records are confidential, patients routinely waive confidentiality in order to provide information to insurers. Patients should be apprised of their right to deny insurers access to their medical records but warned that such a refusal could make it more difficult to obtain insurance coverage later. See Appendix B for more on confidentiality and patients' right to deny access.

The Consensus Panel recommends that clinicians flag charts with positive results, but because of confidentiality concerns, chart reminders should remain neutral and not identify the problem being flagged. Appendix B details three recordkeeping systems that protect patients' privacy.

Negative Screens

Even if the screen is negative, the Consensus Panel recommends periodic rescreening for substance abuse because problematic use of alcohol, illicit drug use, and their consequences can vary over an individual's lifetime. Since there is no clear scientific evidence to define appropriate intervals for screening in asymptomatic patients, the Panel recommends that clinical considerations govern the frequency of rescreening. Indications might include presentation of medical conditions that are often alcohol- or drug-related such as hypertension or insomnia; diabetes or ulcers that do not respond to treatment; persistent requests for prescription drugs; unexplained weight loss; staph infection on face, arms, or legs; frequent falls; repeated fractures, lacerations, or burns; repeated trauma that suggests domestic violence; depression; and sexually transmitted diseases.

Positive Screens

Clinicians should present results of positive screens in a nonthreatening manner. For example, a clinician might say, "After reviewing your answers on the screening questionnaire, there are some things I'd like to follow up with you," or, "Your answers to this questionnaire are similar to the answers of people who may be having a problem with alcohol."

Clinicians must make some quick decisions at the time of screening to determine the appropriate clinical response. Three possible approaches are suggested based on severity of the problem and possible risk (none of the three is appropriate for an intoxicated patient, who may require an immediate response):

1. The clinician can follow up immediately with a brief assessment during the initial visit.
2. The clinician can schedule a subsequent visit for assessment if the screening results are inconclusive.
3. The clinician can decide to refer to another source for assessment.

In areas where specialized substance abuse resources are available, the Consensus Panel recommends that high-risk patients be referred for assessment. The following chapters of this TIP provide information on the next steps: conducting brief assessments and brief interventions and referring and following up on patients who need specialized assessments and treatment.