This Web site is a component of the SAMHSA Health Information Network

Enter Keywords Choose Website This Site Mental Health at HHS All Mental Health

Online Publications Order Publications National Library of Medicine National Academies Press Publications Homepage

printer friendly page e-mail this page bookmark this page shopping cart  current or new account

This Web site is a component of the SAMHSA Health Information Network.

Clinical Preventive Services in Substance Abuse and Mental Health Update: From Science to Services

IX. Adults (19 Years and Older)

Screening and follow-up for tobacco and alcohol use disorders and for depression/anxiety are the primary topics here that are addressed for adults. Each topic is associated with a few brief questions, followed by various low-cost interventions.

In persons without major medical or behavioral comorbidities, the medical benefits of screening and follow-up are substantial but are spread out over too many years to generate immediate health care cost savings. The literature shows tobacco and alcohol quit rates in the range of 5–30 percent (comparing cases to controls) and frequent relapse, but even with these relatively modest quit rates, the benefits are substantial enough to suggest universal implementation. Adult guidelines differ from the adolescent and pregnancy guidelines in that preventing use of illicit drugs is not as urgent an issue.

With depression and anxiety, the shortterm benefit of cost savings depends on the presence or absence of other medical and behavioral comorbidities. In persons without such comorbidities, the benefits are substantial but primarily related to quality of life and workplace productivity. In persons with such comorbidities, reductions in health care costs can be substantial and immediate.

Detailed guidelines and literature reviews are presented for tobacco, alcohol, depression, and anxiety. These guidelines include the specific screening questions to be used, and guidelines for follow-up. A brief discussion is provided relative to substance abuse in adults. Depression-related disorders are presented in detail, with a separate discussion of depression screening, followup, and cost-effectiveness related to heavy users of health care services and those with major chronic diseases.

Separate analyses were done for adults older and younger than 65 years of age. Problems related to use and misuse of multiple prescription drugs in persons over 65 years of age were considered to be outside the scope of this report. Otherwise, preventive guidelines and projected benefits for tobacco, alcohol, and depression screening were so similar for adults older and younger than 65 years of age that the guidelines for all adults are presented in a single section of this monograph for adults 19 years of age and older.

A. Tobacco

After reaching 21 years of age, initiation of tobacco use is rarely a problem. For those who already smoke, the issue to be addressed is prevention of future tobaccorelated medical illness through reduction or elimination of current tobacco use. In randomized trials, inexpensive clinical interventions for cessation of tobacco use have shown increases in abstinence rates in the range of 3–25 percent at one year, compared with controls (Wilson, Taylor, Gilbert, et al., 1988; Okene, Kristeller, Goldberg, et al., 1991; Bronson, Flynn, Solomon, et al., 1989; Hollis, Lichtenstein, Vogt, Stevens, & Biglan, 1993; Cohen, Stookey, Katz, et al., 1989; Curry, Marlatt, Gordon, et al., 1988; Stevens & Hollis, 1989). The key to success is consistent reinforcement by clinicians.

Tobacco-related disease is prevalent enough and serious enough that even with limited efficacy of clinical interventions, such interventions have been deemed worthy of universal implementation. In fact, all major health organizations that have addressed this topic have supported this recommendation. As a result of these findings, tobacco cessation programming for adults and seniors is classified as “general.”

This literature search uncovered no literature dealing directly with the issue of cost-effectiveness for tobacco control programming for adults and seniors. The problem here is two-fold. First, the benefits for adults and seniors without major chronic diseases are too far in the future, and their health care service utilization is too small for smoking cessation to substantially reduce other health care costs within 12 months. Second, for those with major chronic diseases, the major impact is likely to be a reduction in short-term mortality. This reduction in mortality is a substantial patient benefit, but it may increase health care costs by keeping these sick patients alive longer, thereby nullifying the savings from marginal reductions in health care use by these same persons with chronic diseases.

This literature review found no direct evidence that tobacco use influenced a patient’s ability and willingness to follow prescribed regimens of care. The near-term reductions in health care costs attributable to addressing alcohol use, use of illicit substances, and depression do not appear to be a benefit of tobacco cessation programming. As a result, from a healthcare –cost perspective, smoking cessation will have little or no immediate impact on aggregate health care costs.

As tobacco use becomes less and less socially acceptable, fewer people will smoke, and those who do smoke will smoke less. One suggested way to pursue this objective, on a societal basis, is to have every adult and senior asked about tobacco use at every primary care visit, and to have every tobacco user briefly counseled to quit. Although this approach is not amenable to randomized controlled trials, it seems reasonable to presume that action along this line could play a significant role in reducing adult and senior tobacco use.

Interventions
The literature suggests that the topic of tobacco use should be brought up at every outpatient visit. Those who smoke or otherwise use tobacco products may be counseled to cease such use. If immediate cessation seems out of reach, smokers may be counseled to reduce the amount of tobacco they use and to consider enrollment in tobacco cessation programming.

If not already accomplished, steps can be taken to assure that all health and medical facilities are totally smoke-free. This is important for a number of reasons, the most important of which may be communicating to staff and patients that smoking is simply not acceptable because of its extreme hazard to the health of both smokers and persons exposed to secondhand tobacco smoke.

Review of Literature
Evidence Base for Intervention
According to the 1996 Second Edition of the U.S. Preventive Services Task Force’s Guide to Clinical Preventive Services (USPSTF, 1996)—

… Detailed reviews of the extensive literature on the health effects of smoking, dose-response relationships, and nicotine addiction have been published elsewhere (CDC, 1993a; DHHS, 1986, 1989, 1990b; EPA, 1992; NCI, 1993). A number of consistent findings from this body of evidence are well established. First, tobacco is one of the most potent of human carcinogens, causing an estimated 148,000 deaths among smokers annually due to smoking-related cancers (CDC, 1993a). The majority of all cancers of the lung, trachea, bronchus, larynx, pharynx, oral cavity, and esophagus are attributable to the use of smoked or smokeless tobacco (DHHS, 1986, 1989). Smoking also accounts for a significant but smaller proportion of cancers of the pancreas (CDC, 1990; Ghadirian, Simard, & Baillargeon, 1991; Howe et al., 1991; Bueno de Mesquita et al., 1991), kidney (DHHS, 1989), bladder (CDC, 1990; Hartge et al., 1993), and cervix (CDC, 1990; Coker et al., 1992; Sood, 1991; Gram et al., 1992)… . 100,000 deaths from coronary heart disease … [and] 85,000 deaths from pulmonary diseases … .

There is a large body of evidence from prospective cohort and casecontrolled studies showing that many of these health risks can be reduced by smoking cessation (CDC, 1990). A number of clinical trials have demonstrated the effectiveness of certain forms of clinician counseling (Wilson et al., 1988; Okene et al., 1991; Bronson et al., 1989; Hollis et al., 1993; Kottke, Battista, DeFriese, et al., 1988; Cohen et al., 1989) and group (Kottke et al., 1988; Curry et al., 1988; Stevens & Hollis, 1989) in changing the smoking behavior of patients… . A meta-analysis of 39 clinical trials in nonpregnant adults examined different types of clinical smoking cessation techniques involving various combinations of counseling, distribution of literature, and nicotine replacement therapy. It found higher cessation rates in the intervention group compared with the control groups, with differences averaging 6 percent after 1 year (Kottke et al., 1988). Subsequent published trials have demonstrated increases in abstinence rates of 3–7 percent in patients receiving clinician counseling (Wilson et al., 1988; Okene et al., 1991; Bronson et al., 1989; Hollis et al., 1993; Cohen et al., 1989) and 8–25 percent with group counseling, compared with controls (Curry et al., 1988; Stevens & Hollis, 1989). The key elements of effective counseling seem to be providing reinforcement through consistent and repeated advice from a team of providers to stop smoking, setting a specific “quit date,” and scheduling follow-up contacts or visits. Using additional modalities, such as self-help materials, referral to group counseling, advice from more than one clinician, or chart reminders identifying patients who smoke, seems to further enhance effectiveness (Kottke et al., 1988; Cohen et al., 1989; Russell, Wilson, Taylor, et al., 1979; Janz, Becker, Kirscht, et al., 1987; Sanders, Fowler, Mant, et al., 1989).

As adjuncts to counseling, the prescription of nicotine products can facilitate smoking cessation (Lam, Sze, Sacks, et al., 1987; Jarvis, Raw, Russell, et al., 1982; Jackson, Stapleton, Russell, et al., 1986; Tonnesen, Fryd, Hansen, et al., 1988; Hughes, Gust, Kennan, et al., 1989; Tonnesen, Norregaard, Simonsen, et al., 1991; Stapleton, Russell, Feyerabend, et al., 1995; Transdermal Nicotine Study Group, 1991; Daughton, Heatly, Prendergast, et al., 1991; Muller, Abelin, Ehrsam, et al., 1990; Sachs, Sawe, & Leischow, 1993; Fiore, Kenford, Jorenby, et al., 1994; Hurt, Dale, Fredrickson, et al., 1994; Fiore, Smith, Jorenby, et al., 1994). Randomized controlled trials have found that 12-month cessation rates after brief clinician counseling and multiple follow-up visits double from 4 percent to 9 percent with placebo to 9 percent to 25 percent with the nicotine patch (Tonnesen et al., 1991; Stapleton et al., 1995; Sachs et al., 1993; Hurt et al., 1994). When used correctly and in combination with clinician advice to stop smoking, nicotine gum increases long-term smoking cessation rates by approximately one third (Oster, Huse, Delea, & Colditz, 1986; Tang, Law, & Wald, 1994)… . Two metaanalyses of controlled trials of nicotine replacement therapies found a significant benefit for all modalities with no modality being significantly better than another (Tang et al., 1994; Silagy, Mant, Fowler, et al., 1994)… . The evidence suggests that nicotine products are most effective as adjuncts to ongoing smoking cessation counseling (Silagy et al., 1994; Fiore, Jorenby, Baker, et al., 1992). Furthermore, patients need proper instruction on how to use the nicotine replacement therapies. Patients have been reported to use nicotine patches and gum without discontinuing smoking, thus increasing the risk of nicotine toxicity (Johnson, Steven, Hollis, et al., 1992; Orleans, Resch, Noll, et al., 1994).

All major health care organizations and authorities recommend routine clinician counseling of adults, pregnant women, parents, and adolescents to avoid or discontinue smoking and use of smokeless tobacco (USPSTF, 1996; American College of Physicians Health and Public Policy Committee, 1986; AAFP, 1994; AAP, 1988, 1994; ACOG, 1993; Manley et al., 1991; AMA, 1993, 1994a; ADA, 1992; CTF on the Periodic Health Examination, 1994b; NIH, 1989, 1994; American Academy of Otolaryngology—Head and Neck Surgery, 1992; Green, ed., 1994).

The effects on patients with coronary heart disease quitting smoking was reviewed by Critchley and Capewell (Critchley & Capewell, 2003). In this 2003 literature review, they concluded that “quitting smoking is associated with a substantial reduction in risk of all-cause mortality among patients with coronary heart disease. This risk reduction appears to be consistent regardless of age, sex, index cardiac event, country, and year of study commencement.” Thus there is a strong evidence base for a modest reduction in tobacco use through clinician counseling to encourage cessation. The evidence suggests that the health risks of continuing an established tobacco addiction are so extreme, however, that programming of even minimal effectiveness would reap considerable benefits when offered as routine clinical practice.

Nurse-assisted counseling for smokers may be considered by health care systems that provide primary care services in large clinic settings. In 1993, Kaiser Permanente (Portland, Oregon) published a randomized controlled trial (Hollis et al., 1993) showing 86 percent physician participation in delivering brief advice, and quit rates of approximately 7 percent in nurse-counseled patients, compared with approximately 3.9 percent for physician advice alone at one year. Prior to the study, physicians participated in a 1-hour training session to encourage them to use their own words to deliver a basic message lasting no more than 30 seconds:

The best thing you can do for your health is to stop smoking, and I want to advise you to stop as soon as possible. I know it can be very hard; many people try several times before they finally make it. You may or may not want to stop now, but I want you to talk briefly with our health counselor, who has some tips to make stopping easier when you decide the time is right.

The nurse counseling session included a 10-minute video and an assortment of aids and stop-smoking literature. There were three different study interventions— individual, group, and combination—all with similar quit rates (Hollis et al., 1993). The study was limited, given that only about half of the participating cases and controls provided saliva samples for the follow-up testing. Those results were still highly statistically significant, but with results 30–50 percent lower than noted above, if all those who did not submit saliva samples were counted as continuing smokers.

Program Implementation Issues
According to the 1996 Second Edition of the U.S. Preventive Services Task Force’s Guide to Clinical Preventive Services (USPSTF, 1996)—

Although the significant health hazard of tobacco use and the benefits of cessation are well established, studies suggest that many clinicians fail to counsel patients who smoke to stop tobacco use (CDC, 1993c; Anda, Remington, Sienko, et al., 1987; Frankowski & Secker-Walker, 1989; USPSTF, 1996). This reluctance to intervene may be the result of a number of variables, including lack of confidence in the ability to provide adequate counseling, lack of patient interest, lack of financial reimbursement or personal reward, insufficient time, and inadequate staff support (Kottke, Willms, Solberg, et al., 1994). As described above, however, a number of studies have shown that clinician counseling can change behavior, even when the intervention is relatively brief. Nearly 50 percent of all living individuals who have ever smoked have stopped (CDC, 1994a), and 30 percent of quitters report being urged to quit by a physician (Fiore, Novotny, Peirce, et al., 1990). Approximately 90 percent of successful quitters have quit without intensive counseling but by stopping abruptly or with the help of quitting manuals (Fiore et al., 1990). A cost-effectiveness study supports the clinical value of offering smoking cessation counseling during the routine office visit of patients who smoke (Cummings, Rubin, & Oster, 1989).

Certain strategies can increase the effectiveness of counseling against tobacco use (NIH, 1986, 1989, 1994; AMA, 1994a; AAFP, 1987; Kenford et al., 1994):

• Direct, face-to-face advice and suggestions
• Reinforcement
• Office reminders to the physician
• Self-help materials
• Community programs for additional help in quitting
• Drug therapy (nicotine patch or gum and related products)

One recent study (McAfee, Grossman, Dacey, & McClure, 2002) suggested that in a managed care setting (Group Health Cooperative, Tukwila, Washington), a quality improvement initiative using an automated billing system with performance feedback and senior-level incentives could dramatically increase tobacco-related counseling and frequency of intervention, as well as secure the data through the billing system.

Data To Be Gathered
Refer to the section in this monograph called Procedures for Implemention and Evaluation of Preventive Services. Special data issues relative to tobacco and adults are limited to special attention to tobacco use in patients with major chronic diseases. Although not caused by tobacco use, diabetes carries a much higher rate of major complications in smokers.

B. Alcohol

Alcohol screening for all adults, including college students, is classified as “general” because of the severity of both immediate and long-term harms caused by alcohol use by adults. Since both acute use and immediate problems are most severe among college students and other college-age young adults, special attention is directed to this age group (18–29 years of age). The efficacy of clinical interventions to reduce harmful alcohol use is modest, but the severity of the harm—both short-term and long-term— mandates that health care providers and health care delivery systems do what they can to reduce such harmful use. With tobacco and illicit drugs, any use is harmful. With alcohol, however, moderate use can have a favorable effect on all-cause death rates, and on death rates from coronary heart disease (Bradley, Donovan, & Larson, 1993; Stampfer, Rimm, & Walsh, 1993; Maclure, 1993; Klatsky, Armstrong, & Friedman, 1990; Stampfer, Golditz, Willett, et al., 1988; Gaziano, Buring, Breslow, et al., 1993).

There is relatively little clinicians can do to prevent initial excessive use of alcohol, but available, reasonably inexpensive interventions can significantly reduce future excessive use and the behavioral, social, and injury-related complications of such use. In persons with one or more chronic diseases, reducing excessive alcohol use may be of value in improving patient adherence to prescribed regimens of care and avoiding medical complications of excessive alcohol use.

From a primary care perspective, alcoholrelated problems can be divided into two major categories: alcohol dependence/addiction and nondependent problem drinking. The research suggests that those with addiction/dependence should be referred for specialized care. The primary care physician, however, often can successfully manage the nondependent problem drinkers. Separate sections of this report address use of alcohol by pregnant women and by adolescents.

Suggested Interventions
As with tobacco use, the topic of alcohol use may be brought up at every outpatient visit, with follow-up counseling as needed. Unlike tobacco, there appears to be no harm, and some benefit, from one or two drinks per day. This benefit, however, is not substantial enough to recommend that nondrinkers begin to consume alcohol. Practitioners may be careful not to communicate the benefits of moderate use as an excuse for more substantial consumption of alcohol.

Special Service-Related Issues Specific to Adults and Alcohol

• High-quality, validated screening questionnaires that are brief enough to be practical in primary care settings are available for screening adolescents and adults for problem drinking. Adults should be periodically screened for problem drinking or alcohol dependence. In most primary care settings, the twoquestion/ two-item conjoint screen (TICS) or four-question CAGE or CUGE screening instruments may be most useful. In emergency room and psychiatric inpatient settings, the CAGE (four yes/no questions), Audit (10 multiple-choice questions), or Michigan Alcoholism Screening Test (MAST) (Selzer, 1971) (25 questions) may be considered. These are all described below. In community health centers and facility-based primary care outpatient settings that allow nurses or social workers to conduct initial patient settings, use of the 10-question Adult Use Disorders Identification Test (AUDIT) instrument may be considered seriously.
• Clinicians must be able to differentiate problem drinking from alcohol dependence. Problem drinking usually can be successfully managed by the primary care practitioner. Alcohol dependence requires much more intensive intervention, and either specialized programming or specialized health care staff.
• For nondependent problem drinkers, research suggests that the most effective and most well-documented primary care intervention is the Trial for Early Alcohol Treatment (TrEAT) protocol. This involves a defined set of materials and two physician-patient sessions of 10 to 20 minutes apiece. The evidence for this protocol and against single-visit and shorter protocols is described below.
• Unlike tobacco and illicit drugs, modest use of alcohol can have health benefits, such as reducing the risk of heart disease.

Review of Literature
Additional alcohol-and-health literature is presented in the sections of this report addressing the needs of pregnant women and adolescents.

Evidence Base for Intervention
Burden of Suffering
According to the 2003 National Institute on Alcohol Abuse and Alcoholism (NIAAA) health practitioner’s guide to helping patients with alcohol problems (NIAAA, 2003)—

Alcohol problems are common: 14 million American adults suffer from alcohol abuse or alcoholism (Grant, Harford, Dawson, et al., 1994), and more than 100,000 people die from alcohol-related diseases and injuries each year (Stinson, Nephew, Dufour, & Grant, 1996). About a third of all adults engage in some kind of risky drinking behavior, ranging from occasional to daily heavy drinking (NIAAA, 2002). Over the past few generations, patterns of alcohol consumption have changed notably: people start drinking at increasingly earlier ages, the likelihood of dependence has risen in drinkers, and women’s drinking patterns and rates of dependence have become increasingly similar to men’s (Grant, 1997).

According to the 1996 Second Edition of the U.S. Preventive Services Task Force’s Guide to Clinical Preventive Services (USPSTF, 1996)—

Over half a million Americans are under treatment for alcoholism, but there is growing recognition that alcoholism (i.e., alcohol dependence) represents only one end of the spectrum of “problem drinking” (IOM, 1990). Many problem drinkers have medical or social problems attributable to alcohol (i.e., alcohol abuse or “harmful drinking”) without typical signs of dependence (APA, 1994; WHO, 1992), and other asymptomatic drinkers are at risk for future problems due to chronic heavy alcohol consumption or frequent binges (i.e., “hazardous drinking”). Heavy drinking (more than five drinks per day, five times per week) is reported by 10 percent of adult men and 2 percent of women (SAMHSA, 1994). In large community surveys using detailed interviews (Helzer & McEvoy, 1991; Grant et al., 1994; Kessler et al., 1994), the prevalence of alcohol abuse and dependence in the previous year among men was 17–24 percent among 18–29 year-olds, 11–14 percent among 30–44 yearolds, 6–8 percent among 45–64 yearolds, and 1–3 percent for men over 65; among women in the corresponding age groups, prevalence of abuse or dependence was 4–10 percent, 2–4 percent, 1–2 percent, and less than 1 percent, respectively. Problem drinking is even more common among patients seen in the primary care setting (8–20 percent) (Bradley, 1994).

Medical problems due to alcohol dependence include alcohol withdrawal syndrome, psychosis, hepatitis, cirrhosis, pancreatitis, thiamine deficiency, neuropathy, dementia, and cardiomyopathy (NIAAA, 1993). Nondependent heavy drinkers, however, account for the majority of alcohol-related morbidity and mortality in the general population (IOM, 1990). There is a dose-response relationship between daily alcohol consumption and elevations in blood pressure and risk of cirrhosis, hemorrhagic stroke, and cancers of the oropharynx, larynx, esophagus, and liver (Klatsky, Armstrong, & Friedman, 1992; Boffetta & Garfinkel, 1990; Anderson, Cremona, Paton, et al., 1993). A number of studies have reported a modest increase in breast cancer among women drinking two drinks per day or more, but a causal connection has not yet been proven (Rosenberg, Metzger, & Palmer, 1993). Three large cohort studies, involving more than 500,000 men and women, observed increasing allcause mortality beginning at four drinks per day in men (Klatsky et al., 1992; Boffetta & Garfinkel, 1990) and above two drinks per day in women (Fuchs, Stampfer, Colditz, et al., 1995). Women achieve higher blood alcohol levels than do men, due to their smaller size and slower metabolism (Klatsky et al., 1992; Fuchs et al., 1995). Compared to nondrinkers and light drinkers, overall mortality was 30 percent to 38 percent higher among men, and more than doubled among women who drank six or more drinks per day (Klatsky, et al., 1992; Boffetta & Garfinkel, 1990). Of the more than 100,000 deaths attributed to alcohol annually, nearly half are due to unintentional and intentional injuries (CDC, 1990), including 44 percent of all traffic fatalities in 1993 (National Highway Traffic Safety Administration, 1994) and a substantial proportion of deaths from fires, drownings, homicides, and suicides …

The social consequences of problem drinking are often as damaging as the direct medical consequences. Nearly 20 percent of drinkers report problems with friends, family, work, or police due to drinking (NIAAA, 1993). Persons who abuse alcohol have a higher risk of divorce, depression, suicide, domestic violence, unemployment, and poverty (NIAAA, 1993). Intoxication may lead to unsafe sexual behavior that increases the risk of sexually transmitted diseases, including human immunodeficiency virus (HIV). Finally, an estimated 27 million American children are at risk for abnormal psychosocial development due to the abuse of alcohol by their parents (Sher, ed., 1991).

Moderate alcohol consumption has favorable effects on the risk of coronary heart disease (CHD) (Bradley et al., 1993; Stampfer et al., 1988, 1993; Maclure, 1993; Klatsky et al., 1990; Gaziano et al., 1993). CHD incidence and mortality rates are 20 percent to 40 percent lower in men and women who drink one to two drinks/day than in nondrinkers (Fuchs et al., 1995; Klatsky et al., 1990; Stampfer et al., 1988). A meta-analysis of epidemiologic studies suggests little additional benefit of drinking more than 0.5 drinks per day (Maclure, 1993). The exact mechanism for the protective effect of alcohol is not known but may involve increases in high-density lipoprotein (Gaziano et al., 1993) and/or fibrinolytic mediators (Ridker, Vaughan, Stampfer, et al., 1994).

In an update published in 2002, Naimi et al. (2002) noted that nationwide, binge drinking increased from 1993 to 2001. Binge drinking episodes among U.S. adults increased from 1.2 billion to 1.5 billion (25 percent increase), while binge-drinking episodes per person increased by 17 percent, from 6.3 percent to 7.4 percent. Men accounted for 81 percent of binge drinking episodes. Rates of binge drinking episodes were highest among those aged 18–25 years. Binge drinkers were 14 times more likely to drive while impaired by alcohol compared with nonbinge drinkers. There were substantial State and regional differences in per capita binge drinking.

Brief Summary of Available Alcohol Screening Tests for Use in Primary Care Settings

There are a number of screening tests available, ranging from 1 to 25 questions in length, and with substantial variation in sensitivity, specificity, and staff training required for optimal use. None is perfect, but all are better than no screening at all. All of these questionnaire instruments are for screening, not diagnosis. Positive responses appear best when followed up with more extensive interview to confirm or deny the presence of an alcohol-related problem and to differentiate between alcoholism and nondependent problem drinking. These are all described in greater detail in the following section, with sample questions provided.

Single Question:
“On any single occasion during the past 3 months, have you had more than five drinks containing alcohol?”

Two-Question:
“In the last year, have you ever drank or used drugs more than you meant to?” and “Have you felt you wanted or needed to cut down on your drinking or drug use in the last year?”

Four-Question: “CAGE”

C: “Have your ever felt you ought to Cut down on drinking?”

A: “Have people Annoyed you by criticizing your drinking?”

G: “Have you ever felt bad or Guilty about your drinking?”

E: “Have you ever had a drink first thing in the morning to steady your nerves or get rid of a hangover (Eye opener)?”

Four-Question: “CUGE”
The CUGE questionnaire replaces the “annoyed you by criticizing your drinking” question with “Have you often driven under the influence?”

Ten-Question: “AUDIT”(Alcohol Use Disorders Identification Test)

1. How often do you have a drink containing alcohol?
2. How many drinks containing alcohol do you have on a typical day when you are drinking?
3. How often do you have six or more drinks on one occasion?
   (Interviewers are then instructed to skip to questions 9 and 10 if the answer to question 2 is fewer than three drinks, and if the answer to question 3 is “never.”)
4. How often during the last year have you found that you were not able to stop drinking once you had started?
5. How often during the last year have you failed to do what was normally expected from you because of drinking?
6. How often during the last year have you needed a first drink in the morning to get yourself going after a heavy drinking session?
7. How often during the last year have you had a feeling of guilt or remorse after drinking?
8. How often during the last year have you been unable to remember what happened the night before because you had been drinking?
9. Have you or someone else been injured as a result of your drinking?
10. Has a relative, or friend, or doctor, or another health worker been concerned about your drinking, or suggested you cut down?

25-Question: “MAST”
The 25-question Michigan Alcoholism Screening Test (MAST) is relatively sensitive and specific, but it generally is considered too lengthy for routine screening in primary care settings. It is commonly used in psychiatric outpatient and inpatient settings.

Most studies seem to recommend the fourquestion CAGE and CUGE questionnaires for primary care settings, with a minimum of paraprofessional support and use of the 10- question AUDIT questionnaire where nonphysician staff are available to administer and score the questionnaire. The CAGE and CUGE questionnaires require only yes/no answers and are easily memorized by primary care practitioners. The 10-question AUDIT questionnaire has multiple choice questions and a formalized scoring procedure.

More Detailed Discussion of the Accuracy and Utility of Alcohol Screening Tests
According to the 1996 Second Edition of the U.S. Preventive Services Task Force’s Guide to Clinical Preventive Services (USPSTF, 1996)—

Laboratory tests generally are insensitive and nonspecific for problem drinking in both adolescents and adults.

Accurately assessing patients for drinking problems during the routine clinical encounter is difficult. The diagnostic standard for alcohol dependence or abuse (Diagnostic and Statistical Manual of Mental Disorders [DSM] IV) (APA, 1994) requires a detailed interview and is not feasible for routine screening. Physical findings … are only late manifestations of prolonged, heavy alcohol abuse (Glaze & Coggan, 1987). Asking the patient about the quantity and frequency of alcohol use is an essential component of assessing drinking problems, but it is not sufficiently sensitive or specific by itself for screening. In one study, drinking 12 or more drinks a week was specific (92 percent) but insensitive (50 percent) for patients meeting DSM criteria for an active drinking disorder (Buchsbaum, Welsh, Buchanan, et al., 1995). The reliability of patient report is highly variable and dependent on the patient, the clinician, and individual circumstances. Heavy drinkers may underestimate the amount they drink because of denial, forgetfulness, or fear of the consequences of being diagnosed with a drinking problem.

A variety of screening questionnaires have been developed which focus on consequences of drinking and perceptions of drinking behavior. The 25-question Michigan Alcoholism Screening Test (MAST) (Selzer, 1971) is relatively sensitive and specific for DSM-diagnosed alcohol abuse or dependence (84 percent to 100 percent and 87 percent to 95 percent, respectively) (Selzer, 1971; Pokorny, Miller, & Kaplan, 1972), but it is too lengthy for routine screening. . . . The fourquestion CAGE instrument is the most popular screening test for use in primary care (Ewing, 1984), and has good sensitivity and specificity for alcohol abuse or dependence (74 percent to 89 percent and 79 percent to 95 percent, respectively) in both inpatients (Bernadt, Mumford, Taylor, et al., 1982; Bush, Shaw, Cleary, et al., 1987) and outpatients (King, 1986; Buchsbaum, Buchanan, Centor, et al., 1991; Chan, Pristach, & Welte, 1994).

The CAGE is less sensitive for early problem drinking or heavy drinking (Chan et al., 1994; Hays & Spickard, 1987). Both the CAGE and MAST questionnaires share important limitations as screening instruments in the primary care setting: an emphasis on symptoms of dependence rather than early drinking problems, lack of information on level and pattern of alcohol use, and failure to distinguish current from lifetime problems (Chan, Pristach, Welte, et al., 1993).

Some of these weaknesses are addressed by . . . AUDIT, a 10-item screening instrument developed by the World Health Organization (WHO) in conjunction with an international intervention trial. The AUDIT incorporates questions about drinking quantity, frequency, and binge behavior along with questions about consequences of drinking (Saunders, Aasland, Babor, et al., 1993). . . . AUDIT had high sensitivity and specificity for “harmful and hazardous drinking” (92 percent and 94 percent, respectively) as assessed by more extensive interview (Saunders et al., 1993). . . . Because it focuses on drinking in the previous year, however, AUDIT is less sensitive for past drinking problems (Schmidt, Barry, & Fleming, 1995).

The World Health Organization (WHO) recommends use of the AUDIT questionnaire in all primary care settings. A guidelines document for use with AUDIT is available free of charge from WHO (Babor TF, Higgins-Biddle, & Monterio, 2001). The questionnaire consists of 10 questions dealing with consumption of alcohol, symptoms of dependence, and social and behavioral evidence of harm. Each of the questions is scored on a scale of 0–4, with scores of 16–19 warranting supplemental counseling and continued monitoring. Scores in the range of 20–40 suggest referral to a specialist for diagnostic evaluation and treatment.

The AUDIT and MAST questionnaires require written forms and formal scoring procedures, which in turn require more staff training. These are not problems with the shorter and simpler TICS, CAGE, and CUGE, which are short enough to be easily memorized by the primary care physicians and nurses and elicit yes/no answers.

Use of Screening Tests for Alcohol Problems
Numerous studies demonstrate that clinicians frequently are unaware of problem drinking by their patients (USPSTF, 1996; NIAAA, 1993; Weisner & Matzger, 2003). Early detection and intervention may alleviate ongoing medical and social problems resulting from drinking and reduce future risks from alcohol abuse (USPSTF, 1996).

In 1998, the Substance Abuse Task Force of the Society for Academic Emergency Medicine issued a statement urging emergency room physicians to use screening questionnaires to improve their detection of alcohol-related problems in the emergency department setting. The Task Force asserted that early detection of alcohol problems would provide an opportunity for early intervention, which in turn might reduce subsequent morbidity and mortality (D’Ononfrio et al., 1998).

In a review of the quality of health care provided to adults in the United States, published in the New England Journal of Medicine in 2003, McGlynn et al. reported that, of all quality measures explored, adherence to quality measures for alcohol dependence was documented in only 10.5 percent of records reviewed. This compares with approximately 40 percent to 78 percent for most other quality measures in this study (McGlynn et al., 2003). This study gathered data from adult surveys and medical record reviews in 12 metropolitan areas of the United States for the most recent 2-year period.

Several recent papers urged screening of patients with depression (Abraham & Fava, 1999), schizophrenia (Agelink, Ullrich, Lemmer, Dirkes-Kersting, & Zeit T, 1999) and/or mood, anxiety, and substance use disorder (DeGraff, Bijl, Smit, Vollenbergh, & Spijker, 2002) for alcohol-related problems, given the high prevalence of comorbid alcohol problems in these patients as well as self-medication with alcohol.

A brief overview of screening tests for alcohol problems can be found on the Web site of the National Institute on Alcohol Abuse and Alcoholism (NIAAA) at http://pubs.niaaa.nih.gov/publications/aa56.htm. This April 2002 review (NIAAA, 2002) makes the following major points:

• Both questionnaires and blood tests are available. The blood tests (GGT [Gamma-glutamyl transferase]; CDT Special Report 78 [Carbohydrate Deficient Transferrin]; MCV [Mean Corpuscular Volume]; and possibly FAEEs [Fatty Acid Ethel Ethers]) probably are of little value in screening for chronic alcohol problems, but may be of significant value in tracking the progress of alcoholics and problem drinkers under care.
• The screening tests are not diagnostic. They identify individuals who may be interviewed more carefully to confirm or deny the impression of an alcohol-related problem, before establishing the need for further investigation, treatment, or referral.
• Use of screening tests is very effective both in identifying individuals with alcohol-related problems, and getting them the appropriate therapy (Fiellin, Reid, & O’Connor, 2000).
• The CAGE questionnaire (Ewing, 1984) has been verified extensively, with sensitivities for detecting alcohol abuse and alcoholism (Fiellin et al., 2000) ranging from 43 to 94 percent. It is well suited to primary care practice because it poses four straightforward yes/no questions that the clinician can easily remember, and it requires less than a minute to complete. This test, however, may fail to detect low but risky levels of drinking (Fiellin et al., 2000), and often performs less well among women and socially vulnerable populations (Cherpitel, 1999; Steinbauer et al., 1998).
• The performance of CAGE can be improved by incorporating questions about the quantity and frequency of drinking, as recommended by NIAAA in The Physicians Guide to Helping Patients With Alcohol Problems (NIAAA, 1995). This approach worked well in a general population sample (Dawson, 2000) and did better than CAGE alone among African Americans in an urban emergency room (Friedman, Saitz, Gogineni, Zhang, & Stein, 2001).
• The Alcohol Use Disorders Identification Test (AUDIT) (Saunders et al., 1993) also incorporates questions about quantity and frequency of alcohol use. In contrast to CAGE, AUDIT compares favorably with other instruments in detecting risky drinking but is less effective in identifying alcohol use and alcoholism (Fiellin et al., 2000; Reinert & Allen, 2002). AUDIT has proven useful among medical and psychiatric inpatients, in emergency rooms (Reinert & Allen, 2002), and in the workplace (Reinert & Allen, 2002; Hermansson, Helander, Huss, Brandt, & Ronnberg, 2000; Hermansson, Helander, Brandt, Huss, & Ronnberg, 2002). AUDIT is relatively free of gender and cultural bias (Cherpitel, 1999; Reinert & Allen, 2002; Volk et al., 1997). In addition, it shows promise for screening adolescents and older people, populations in which standard screening instruments produce inconsistent results (Steinbauer et al., 1998; Reinert & Allen, 2002; Clay, 1997; Chung et al., 2000; 2002). The major disadvantages of AUDIT are its length (10 questions) and relative complexity (multiple choice); clinicians require training to score and interpret the test results (Allen & Columbus, 1995).
• Alcohol consumption puts people at greater risk of injury. It plays a role in a large percentage of trauma incidents, including motor vehicle crashes. RAPS4 is a four-item questionnaire derived in part from TWEAK and AUDIT. In both primary care and emergency room settings, RAPS4 showed consistently high sensitivity for detecting alcoholism across gender and ethnic subgroups, although its utility for screening for risky drinking or alcohol abuse has yet to be proven (Cherpitel, 2000; Borges & Cherpitel, 2001).

More information on these and other alcohol-related screening tests also can be found on the NIAAA Web site at www.niaaa.nih.gov.

In a study published in 2000, Aertgeerts (Aertgeerts et al., 2000), working from Catholic University in Belgium, compared several screening questionnaires in a population of 3,564 consecutive college freshman and concluded that a modified CAGE questionnaire, which is called “CUGE,” may improve screening in college students. The CUGE questionnaire replaces the “annoyed you by criticizing your drinking” question with “often driving under the influence.”

A series of four recent papers (Williams & Vinson, 2001; Taj, Devera-Sales, & Vinson, 1998; Aertgeerts, Buntinx, Ansoms, & Fevery, 2001; Seppa, Lepisto, & Sillanaukee, 1998) reported that variants on the theme of a single “five-shot” question generated results comparable to CAGE and AUDIT in adult male and female patients. The basic question was: “On any single occasion during the past 3 months, have you had more than five drinks containing alcohol?” Perhaps the most reasonable interpretation is that of Taj et al. (1998): “A single question about alcohol can detect at-risk drinking and current alcohol-use disorders with clinically useful positive and negative predictive values.”

Another recent study (Brown, Leonard, Saunders, & Papasouliotis, 2001) presented promising, but as yet unverified results for a two-question questionnaire—the two-item conjoint screen (TICS) for alcohol and other drug abuse that can be incorporated easily into routine clinical practice. The two questions are: “In the last year, have you ever drunk or used drugs more than you meant to?” and “Have you felt you wanted or needed to cut down on your drinking or drug abuse in the last year?”

Of all studies considered, the four-item CAGE and CUGE questionnaires are probably the most appropriate for most primary care settings. They are detailed earlier in this chapter.

Alcohol Abuse Diagnostic Criteria
The following criteria have been adapted from the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM–IV), published by the American Psychiatric Association. This is as published in the NIAAA Health Practitioner’s Guide (NIAAA, 2003). The criteria are as follows, with one or more of these situations occurring at any time in the past 12 months:

• Failure to fulfill major role obligations at work, school, or home because of recurrent drinking
• Recurrent drinking in hazardous situations
• Recurrent legal problems related to alcohol
• Continued use despite recurrent interpersonal or social problems

Alcohol Dependence Diagnostic Criteria
The following criteria have been adapted from the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM–IV), published by the American Psychiatric Association. This is as published in the NIAA Health Practitioner’s Guide (NIAAA, 2003). The criteria are follows, with three or more of these situations occurring at any time in the past 12 months:

• Tolerance (need to drink more to get the same effect)
• Withdrawal syndrome or drinking to relieve withdrawal
• Impaired control (unable to stop drinking)
• Drank more or longer than intended
• Neglect of activities
• Time spent related to drinking or recovering
• Continued use despite recurrent psychological or physical problems

Effectiveness of “Brief Interventions” for Nondependent Problem Drinkers
Typical of the results for nondependent drinkers, a meta-analysis of six brief intervention trials (5–15 minutes of clinical counseling) showed an average reduction in alcohol consumption of 24 percent, comparing cases to controls. Although selfreported consumption may be subject to bias, reported changes in drinking correlated with measures of GTT and blood pressure in most studies (USPSTF, 1996; Babor et al., 1992). It is important to note, however, that this and most other such studies suffered from important methodological limitations (USPSTF, 1996). Since publication of the 1996 Guide (as quoted above), there have been several publications, which among them appear to bring this issue into clearer focus for nondependent problem drinkers.

In mid-1996, WHO published the results of a randomized, controlled trial of two brief interventions in 1,260 men and 299 women in study centers scattered across 10 countries, including the United States (WHO Brief Intervention Study Group, 1996). The subjects were selected to be nondependent, heavy drinkers. The two interventions tested were a single, 5-minute “simple advice” session and a 20-minute “brief counseling” session, both supported with various written educational materials. Each intervention was delivered in a single session, with patients followed up 9 months later. On interview 9 months later, men reported 17 percent lower average daily alcohol consumption, and women reported a 10 percent decrease. There was no difference between those getting the 5-minute “simple advice” and those receiving the more intensive 20-minute “brief counseling” session. Although promising, weaknesses in the study design raise questions about the firmness of the findings. This WHO study frequently is referenced in newspapers and other nonresearch publications as proof that even the briefest of interventions are of value; however, this conclusion has not been borne out in other studies.

In 1999, Poikolainen published a metaanalysis of brief interventions in problem drinkers comparing single-session “brief interventions” with multi-session “extended brief interventions” (Poikolainen, 1999). His review of the literature did not include the WHO study referenced above because it apparently did not meet his criteria for inclusion in the review on methodological grounds. His review of multiple other publications, including 14 separate datasets, concluded that the single-session brief interventions were of little or no value, and that the multiple-session interventions were clearly beneficial in women, and sometimes but not always beneficial in men.

The best documented and methodologically strongest recent trial is the Project TrEAT (Trial for Early Alcohol Treatment) published by Fleming et al. in a series of papers from 1997 to 2000 (Fleming, Mundt, French, Manwell, Stauffacher, & Barry, 2000; Fleming, Manwell, Barry, Adams, & Stauffacher, 1999; Fleming, Barry, Manwell, Johnson, & London, 1997). This series of papers looked at nondependent problem drinkers in 17 primary care and managed care sites in Wisconsin. There were 382 controls and 392 intervention patients. The intervention consisted of two 10–15 minute counseling sessions by the primary care physician, with written support materials. Patients were followed up at 6 months and 12 months. Depending on the measure, differences in alcohol consumption between cases and controls were in the range of 20–50 percent at 12 months. This difference was significant enough to reduce emergency room and hospital bed use within the first 12 months to more than cover the $205 estimated average per-case cost of the intervention. Considering only health care costs, the benefit-cost ratio was about 2.5:1. If avoided costs of crime and motor vehicle accidents are included, the benefit-cost ratio increases to 5.6:1.

Yet another controlled clinical trial demonstrating the lack of efficacy of singlesession counseling sessions was published in 2000 (Freeborn, Polin, Hollis, & Senft, 2000). This trial, with 514 participants in a managed care setting (Kaiser Permanente, Portland, Oregon) showed a nonsignificant reduction in alcohol consumption at 6 months, but no reduction in health care utilization when comparing cases with controls.

Effectiveness of “Brief Interventions” for Dependent/Addicted Drinkers
For adults with alcohol-dependency completing either inpatient treatment or 12 weeks of outpatient treatment, some studies have shown long-term abstinence rates of approximately 60 percent. These data are difficult to interpret, however, because of inadequate control groups, insufficient or selective follow-up, and selection bias due to the characteristics of patients who successfully complete voluntary treatment programs (USPSTF, 1996; IOM, 1989; Thurstin et al., 1986; Emrick, 1987). Since spontaneous remission occurs in as many as 30 percent of alcoholics (USPSTF, 1996; Smart, 1975/1976; Saunders & Kershaw, 1979), reduced consumption may be inappropriately attributed to treatment. Successful treatment is likely to represent a complex interaction of patient motivation, treatment characteristics, and the posttreatment environment (family support, stress, etc.) (USPSTF, 1996; IOM, 1990; NIAAA, 1993). The IOM review concluded that treatment of other life problems (e.g., with antidepressant medication, family or marital therapy, or stress management) and counseling with empathetic therapists were likely to improve treatment outcomes (USPSTF, 1996; IOM, 1989).

Program Implementation Issues:
How To Manage the Intervention So That It Succeeds in Securing Desired Benefits

• Based on the research, the primary program implementation issue relative to alcohol-related screening and intervention appears to be strict adherence to the details of screening and intervention protocols, especially if the TrEAT protocol is to be used. Given the nature of the protocol, it is all too easy to defer the counseling sessions from the primary care physicians to other staff, or to reduce the content length of the sessions. The literature shows, however, that doing so may substantially reduce, if not eliminate, the benefit to be secured from the intervention.
• The second program implementation issue relative to alcohol-related screening has to do with the structure and staffing of the primary care setting. In settings without adequate nursing and/or health education support staff, the research indicates that it may be better to proceed with one of the simpler one to four question screening instruments than to attempt to use the 10- question AUDIT instrument on a selective or inconsistent basis.
• Finally, special attention can be paid to policies and procedures and staff and physician education to ensure adequate screening and follow-up, and to enable the staff to better differentiate between alcohol dependence/addiction and nondependent problem drinking.

Data To Be Gathered
Refer to Appendix D. Supplemental data needs relative to alcohol and adults include the following:

• Prevalence of alcoholism, cirrhosis, and other specific alcohol-related disorders
• Incidence of alcohol-related injury, suicide, and homicide within the enrolled population
• Alcohol-related utilization of outpatient, inpatient, and emergency services
• Separate tracking of services to address problem drinking and alcohol dependence, with follow-up to prevent and address relapse, and to document the success (or lack thereof) of the programming

C. Adult Use/Abuse of Illicit Drugs

There is remarkably little in the way of published, peer-reviewed literature dealing with the issue of prevention of adult use of illicit drugs. The conventional wisdom appears to be that initiation of illicit drug use is relatively uncommon beyond young adulthood unless such use is self-medication for stress, depression, or another behavioral disorder. There seems to be no need for health care systems to initiate specific programming to prevent initiation of illicit drug use by adults.

Treating adults, especially younger adults, for use of illicit substances is an important therapeutic issue and generally is handled in the emergency room and by mental health professionals, rather than by primary care practitioners. Preventive issues generally are limited to those noted in the following review of pertinent literature.

A related topic is misuse and abuse of prescription medications among adults, especially older adults who have minor depression and/or who use multiple medications to control multiple chronic diseases. This is a serious problem, but because it is more therapeutic than preventive, it is considered beyond the scope of this current literature review.

Intervention
Although clinical management of adult use of illicit drugs is appropriate, no screening or other preventive services are suggested for adults concerning illicit drugs. A partial exception may be the need to counsel older adults about possible abuse of prescription medication. Further discussion of this topic is beyond the scope of this report.

Review of Pertinent Literature
Although none of the studies noted below is a randomized clinical trial, the studies do provide background information on the issue of adult use/abuse of illicit drugs for health care policymakers who wish to further explore this issue.

In a 1998 literature review, Drake et al. (Drake, Mercer-McFadden, Mueser, McHugo, & Bond, 1998) noted that patients with severe mental disorders, such as schizophrenia and co-occurring substance use disorders, frequently receive treatment for their disorders from multiple clinicians in parallel treatment systems. Their review provides promising evidence that integrating the treatment of these patients through a single set of clinicians can yield promising results in terms of remission of the underlying mental disorder, reduction in substance use disorder, and use of health care resources.

Frankin and Hendrix, in an uncontrolled study published in 2001 (Franken & Hendriks, 2001), noted that screening persons with substance use disorder for underlying anxiety and mood disorders using the SCL-90 (Symptom Checklist 90) questionnaire (Franken & Hendricks, 2001) could be of significant value in controlling both disorders and reducing the need for more extensive psychiatric diagnostic evaluation.

Schermer and Wisner, in a record review of patients suffering from major trauma, published in 1999 (Schermer & Wisner, 1999), urged screening of patients suffering from major trauma for methamphetamines and cocaine. This California study noted a doubling of methamphetamine rates from 7.4 percent to 13.4 percent in patients suffering from major trauma from 1989 to 1994; minimal increases in cocaine positivity, from 5.8 percent to 6.2 percent; and a decrease in alcohol positivity, from 43 percent to 35 percent.

In a record review published in 2001, Chitwood et al. (Chitwood, Sanchez, Comerford, & McCoy, 2001), noted that injection drug users, other sustained drug users, and “heavy” alcohol users were less likely to avail themselves of preventive services than other patients being seen in their Miami, Florida, center.

Bennet and Beaudin, in an opinion piece published in 2000 (Bennett & Beaudin, 2000), provided a guide to facilitate collaboration between employers and managed care plans to address substance use disorder in the workplace.

Data To Be Gathered
Since there are no suggested screening or other preventive services, there are no data needs specific to preventive services, illicit drugs, and adults.

D. Depression and Anxiety

Depression-related disorders (generalized anxiety disorder, minor depression, major depression, and bipolar disorder) are common, serious, readily treatable, and more often than not either missed or ignored in primary care settings. Effective and costefficient screening procedures are readily available, but they should be used with caution because of the importance of differentiating between the depressionrelated disorders noted above.

From a preventive perspective, it is important to differentiate major depression from the other listed disorders, and to consider separately the consequences of depression in otherwise healthy adults as compared with adults who have major medical and/or psychiatric comorbidities. In both populations, depression dramatically increases the risk of suicide, dramatically reduces the quality of life, and unfavorably affects workplace productivity. Among those with major chronic diseases, however, depression dramatically reduces the ability and willingness of the patient to adhere to prescribed regimens of care. In this chronic disease group, detection and skilled management of the depression has been shown in research to be cost-effective in terms of other health care costs.

Routine screening for depression among all adult outpatients was given a universal rating by the U.S. Preventive Services Task Force in 2002 (USPSTF, 2002b, 2003). There are no medical means to prevent depression (Munoz, 1993). There are, however, effective means to screen and then manage the depression in a cost-efficient way to improve the quality of life of the patient, reduce other health care costs, and substantially reduce the risk of suicide.

Intervention
Rigorous research demonstrates that all adults should be screened for depression at every outpatient visit. A simple two-question screen is likely to be as effective as longer screening instruments. The two questions are: “Over the past 2 weeks, have you felt down, depressed, or hopeless?” and “Over the past 2 weeks, have you felt little interest or pleasure in doing things?” (USPSTF, 2002b, 2003). These questions are not diagnostic, but they do serve as a starting point for further exploration of depressive symptoms to determine the need for referral to mental health specialists and/or prescription of antidepressant medications.

The literature supports every health care delivery system developing and maintaining the capacity to follow up with more definitive diagnostic interviews and appropriate patient management. Although much of this can be managed with supplemental training of primary care practitioners, it is important to have access to mental health professionals for the more difficult cases, and to properly differentiate anxiety disorders and minor depression from major depression, as well as unipolar depression from bipolar (manic-depressive) disorder.

Summary of 2002 U.S. Preventive Services Task Force Recommendations: Depression
In April 2002, the U.S. Preventive Services Task Force (USPSTF) issued an updated report on depression (Pignone et al., 2002; USPSTF, 2003). The USPSTF is an independent panel of experts in primary care and prevention that systematically reviews the evidence of effectiveness and develops recommendations for clinical preventive services (USPSTF, 2003). These new recommendations have been incorporated into the newly developing Guide to Clinical Preventive Services, 3rd Edition, 2000–2003. This update guide is not yet available in book form, but it is readily accessible on the Internet site of the Agency for Healthcare Quality Research (AHRQ) at www.ahrq.gov.

The best way to access the depression recommendation and evidence base is to: 1) go to the Web site; 2) click on “Clinical Information: Preventive Services,” 3) click on “U.S. Preventive Services Task Force (USPSTF),” 4) click on “Mental Disorders and Substance Abuse,” and 5) click on “Depression: Screening.” This will lead to the summary and full text of the April 2002 literature review. The site and all its reports are available to the public, free of charge, with no requirement for a password or any form of registration.

The following provides a series of quotations from the summary and the literature review, which have been selected to meet the needs of health care system administrators, benefit managers, and fiscal officers. Those desiring more detailed information are urged to access the full recommendations and full literature review on the AHRQ Web site.

Summary of Recommendations

The U.S. Preventive Services Task Force (USPSTF) recommends screening adults for depression in clinical practices that have systems in place to assure accurate diagnosis, effective treatment, and follow-up. [The USPSTF] gave a “B” recommendation [which means] clinicians should routinely provide the service to eligible patients; [there is] at least fair evidence that the service improves important health outcomes and [USPSTF] concludes that benefits outweigh harms.

. . . Trials that have directly evaluated the effect of screening on clinical outcomes have shown mixed results. Small benefits have been observed in studies that simply feed back screening results to clinicians. Larger benefits have been observed in studies in which the communication of screening results is coordinated with effective follow-up and treatment.

The USPSTF concludes that the amount and rigor of research to date are insufficient to recommend for or against routine screening of children or adolescents for depression.

Epidemiology and Clinical Consequences

… In primary care settings, the point prevalence of major depression ranges from 5 to 9 percent among adults, and up to 50 percent of depressed patients are not recognized. Other disabling depressive illnesses (that also are amenable to treatment) include dysthymia (a chronic lowgrade depression) and minor depression (an episodic, less severe illness). These two illnesses are as common as major depression in primary care settings.

Diagnosis of Major Depression

The prevailing standard of the American Psychiatric Association for the diagnosis of depression is the opinion of an examining clinician that a patient’s symptoms meet the criteria described in the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM–IV) (APA, 1994). This creates a situation in which, following the initial screening, there must be further questioning by the primary care practitioner to confirm or deny the impression of possible depression, and differentiate minor from major depression. The diagnosis of major depression is based upon the daily presence of four or more of the following symptoms, along with sadness or apathy, for at least 2 weeks (Dornbrand, Hoole, & Pickard, 1992):

1. Decreased or increased appetite, weight change
2. Insomnia or increased sleeping
3. Observable change in psychomotor activity, either agitation or retardation
4. Persistent inability to enjoy usually pleasurable activities, including sex
5. Fatigue
6. Feelings of worthlessness or guilt
7. Slowed thinking or decreased concentration
8. Recurrent thoughts of death or suicide

Accuracy and Reliability of Screening Tests

… Assuming optimal test performance and a prevalence of major depression of 5–10 percent in primary care settings, approximately 24–40 percent of patients who screen positive will have major depression. Some patients with “false-positive” results on screening may have dysthymia or subsyndromal depressive disorders (depressed, but not depressed enough to meet diagnostic criteria for major depression) that might benefit from treatment or closer monitoring; others may have comorbid disorders such as anxiety disorder, substance abuse, panic disorder, posttraumatic stress disorder, or grief reactions; still others may have no disorder at all. The finding of a positive screen therefore requires further diagnostic questioning by the clinician to establish an appropriate diagnosis and initiate a plan for treatment and follow-up.

Effectiveness of Early Treatment

Effective treatments are available for patients with depressive illness detected in primary care settings. Antidepressant medications for major depression are clearly more effective than placebo. Newer agents (medications) perform similarly to older agents.

Psychosocial and psychotherapeutic interventions are probably as effective as antidepressant medications for major depression, but they are clearly more timeintensive. Few studies have examined the effect of combining medications and psychotherapy.

Effectiveness of Screening

Trials that examined the effect of feedback of screening results on the proportion of depressed patients who received treatment showed mixed results: in four fair-to-good quality trials that used feedback alone, there was no significant effect on treatment rates, but four of the five trials that combined feedback with treatment advice or other systems support reported increased treatment rates in the intervention group.

All three trials that compared the effects of integrated recognition and management programs with usual care in community primary care practices showed significantly improved patient outcomes. Integrated programs included feedback, provider and/or patient education, access to case management and/or behavioral care, telephone follow-up, and institutional commitment to quality improvement.

Potential Harms of Screening and Treatment

The potential harms of screening include false-positive screening results, the inconvenience of further diagnostic workup, the adverse effects and costs of treatment for patients who are incorrectly identified as being depressed, and potential adverse effects of labeling. None of the research reviewed provided useful empirical data regarding these potential adverse effects.

Recent History and Recent USPSTF Recommendation
Much of the expanded interest in depression is due to the advent of better-tolerated antidepressant medications (Olfson et al., 2002) and the cost-effectiveness of screening for depression and managing depression in patients with major medical and psychiatric comorbidities. These factors converged to increase the percentage of adult outpatients treated for depression from 0.73 per 100 in 1987 to 2.23 in 1997. During this same period, the proportion of individuals treated with antidepressant medications increased from 37.3 percent to 74.5 percent. Use of psychotherapy also decreased in these patients from 71.1 percent to 60.2 percent, and the locus of much of this care moved from psychiatrists to primary care practitioners (Olfson et al., 2002).

In presenting their recommendations, the USPSTF suggested the use of a simple twoquestion screen as likely to be as effective as longer screening instruments. The two questions are: “Over the past 2 weeks, have you felt down, depressed, or hopeless?” and “Over the past 2 weeks, have you felt little interest or pleasure in doing things?” (USPSTF, 2002b, 2003). These questions are not diagnostic, but they do serve as a starting point for further exploration of depressive symptoms to determine the need for referral to mental health specialists and/or for prescription of antidepressant medications.

In making these recommendations, the Task Force was careful to specify that such screening should only be done on a routine basis in health care delivery systems with the capacity to follow up with more definitive diagnostic interviews and appropriate patient management. This last caveat was apparently inserted to refer to education of primary care practitioners and access to mental health professionals to properly differentiate anxiety disorders and minor depression from major depression, and unipolar depression from bipolar (manic-depressive) disorder.

The USPSTF recommendation relative to screening for depression is based on caseseries studies showing the ability to detect depression using a variety of screening procedures, as well as the efficacy of treating the cases detected using the screening procedure—with the efficacy of such treatment well documented in randomized controlled trials.

A major practical issue at the interface of the primary care physician and patient is the differentiation of anxiety disorders and minor transient depression from major recurrent depression. Although all these disorders can benefit from counseling, guidance, and antidepressant medication and all affect workplace productivity, major depression is the one with the most substantial impact on health care costs and the highest risk of suicide. Differentiating anxiety disorders and minor depression from major depression also is important because once diagnosed, medication for major depression should be maintained for at least 6 months to prevent current and future relapses. For patients with uncomplicated general anxiety or minor depression, reassurance, counseling, relaxation therapy, and stress management techniques often are effective without medication (Margolis & Swartz, 2002). Major depression generally requires more aggressive treatment.

Differential Diagnosis (From a Management/Policy Perspective)
Generalized Anxiety Disorder
Patients with a problem of “nerves” account for approximately 10–30 percent of encounters in general medical practice. They may complain of being “shaky,” “tense,” “irritable,” or “uptight,” or the diagnosis may be made in the course of evaluating a somatic complaint (Dornbrand et al., 1992). Initial manifestations most commonly present between 20 and 35 years of age, with a slight preponderance in women (Tierney, McPhee, & Papadakis, 2003). Generalized anxiety disorder that presents for the first time after the age of 40 should probably be considered evidence of depression until proven otherwise (Dornbrand et al., 1992).

Anxiety disorders appear to be underrecognized and untreated even though treatment interventions have been shown to be effective and cost-efficient (Rice & Miller, 1998). Unfortunately, there are no verified questionnaire instruments short enough for routine use in primary care settings, as with alcohol use disorders and depression.

Depression
According to the 2002 Systematic Evidence Review, which serves as the basis for the USPSTF depression guideline (Pignone et al., 2002; USPSTF, 2003)—

Burden of Suffering

Depressive disorders are common, chronic, and costly. Lifetime prevalence rates from communitybased surveys range from 4.9 percent to 17.1 percent (Kessler et al., 1994; Robins & Regier, 1991; Depression Guideline Panel, 1993). In primary care settings, the prevalence of major depression is 6–8 percent (Katon, 1987). Longitudinal studies suggest that approximately 80 percent of individuals experiencing a major depressive episode will have at least one more episode during their lifetime, with the rate of recurrence even higher if minor or subthreshold episodes are included (Judd, 1997). Approx-imately 12 percent of patients who experience depression will have a chronic, unremitting course (Judd, 1997). The substantial public health and economic significance of the chronic illness is reflected by the considerable utilization of health care visits and tremendous monetary costs: $43 billion (1990 dollars) annually, with $17 billion of that resulting from lost work days (Greenberg, Stiglin, Finkelstein, & Berndt, 1993).

The burden of suffering from depression is substantial. Suicide, the most severe of depressive sequelae, has a rate of approximately 3.5 percent among all cases with major depression, a risk that increases to approximately 15 percent in people who have required psychiatric hospitalization (Blair-West & Eyeson-Annan, 1997). The specific risk for suicide associated with depressive disorders is elevated 12- to 20-fold compared with the general population (Harris & Barraclough, 1997). The World Health Organization (WHO) identified major depression as the fourth leading cause of worldwide disease burden in 1990, causing more disability than either ischemic heart disease or cerebrovascular disease. Its associated morbidity is expected to increase; unipolar depressive illness is projected to be the second leading cause of disability worldwide in 2020. Furthermore, depression appears to contribute to increased morbidity and mortality from other medical disorders, such as cardiovascular disease (Musselman, Evans, & Nemeroff, 1998).

Both the chronicity and recurrence of depressive illness play a large role in depression’s heavy disease burden. The more severe a depression becomes and the longer it lasts, the greater the likelihood that the depression will become chronic (Consensus Development Panel, 1985). Consequently, early effective identification and management of depressive illness will not only decrease the substantial morbidity associated with the current episode but may also decrease the likelihood that the illness will become chronic, with its additional associated morbidity (Pennix et al., 1998).

According to the 1996 Second Edition of the U.S. Preventive Services Task Force’s Guide to Clinical Preventive Services (USPSTF, 1996)—

Depression is more common in persons who are young, female, single, divorced, separated, seriously ill, or who have a prior history or family history of depression (Weissman, 1987).

Major depressive disorder can result in serious sequelae. The suicide rate in depressed persons is at least eight times higher than that of the general population (Monk, 1987). In 1993, 31,230 suicide deaths were reported, although the actual number is probably much higher (National Center for Health Statistics, 1994). Most persons who commit suicide have a mental disorder, with depression associated with approximately half of suicides (Greenberg et al., 1993; Weissman, 1987). The incidence of documented suicides by adolescents and young adults has dramatically increased in recent decades, with 5,000 youths committing suicide each year and perhaps as many as 500,000–1,000,000 making an attempt (Greydanus, 1986).

On a population basis, the most important effect of major depression may be on quality of life and productivity rather than suicide. This effect is widespread and has been shown to be comparable to that associated with major chronic medical conditions such as diabetes, hypertension, or coronary heart disease (Wells, Stewart, Hayes, et al., 1989; Broadhead, Blazer, George, & Tse, 1990). Also, depressed persons frequently present with a variety of physical symptoms—three times the number of somatic symptoms of controls in one study (Waxman, McCreary, Weinrit, & Carner, 1985). If their depression is not recognized, these patients may be subjected to the risks and costs of unnecessary diagnostic testing and treatment (Katon & Russo, 1989; Katon, Berg, Robins, & Risse, 1986).

The main task of evaluation in primary care settings is to identify the 5–13 percent of patients with the specific psychobiologic disorder—major depression—that will require 6 months of medication and longterm follow-up (Dornbrand et al., 1992). Greenberg et al. estimated the total cost of depression to American society to be approximately $43.7 billion in 1990 (Greenberg et al., 1993). Given the frequent co-occurrence of anxiety and depressive disorders (in which the anxiety would be considered a symptom of the depression), this estimate is reasonably consistent with the estimates provided by DuPont, Rice, and Miller (DuPont et al., 1996; Rice & Miller, 1998) in the preceding discussion of generalized anxiety disorder.

The economic cost of anxiety disorders in the United States is highlighted in two papers by DuPont, Rice, and Miller, one each published in 1996 and 1998 (DuPont et al., 1996; Rice & Miller, 1998). Considering all costs to American society, both medical and nonmedical, they estimated that the total cost of all mental illnesses to American society was $148.8 billion in 1990. Anxiety disorders were estimated to affect more than 10 percent of the U.S. population at some point in their lives, with a total cost of $46.6 billion in 1990. Of this, approximately three quarters were due to lost productivity. This demonstrates that the major economic impact is in the workplace, not in health care costs. Affective disorders, with much of the cost related to depression, cost American society another $30.4 billion in 1990. Given that anxiety can present as a symptom of depression, this group of disorders (anxiety and depression combined) account for more than half of the total cost of mental disorders in the United States.

In 2003, Stewart et al. (Stewart, Ricci, Chee, Hahn, & Morganstein, 2003) published data from a survey of employed individuals who participated in the American Productivity Audit, conducted August 1, 2001, through July 31, 2002. This study was based on 692 persons who responded affirmatively to two depression screening questions, and a stratified random sample of 435 persons who responded in the negative. All of these individuals were then recruited for and completed a supplemental interview.

Extrapolating from this sample, workers with depression lost 5.6 hours per week of health-related productive time, compared with 1.5 hours per week for those without depression. Eighty-one percent of the time lost was due to reduced performance while at work. Major depression accounted for 48 percent of the lost productive time among those with depression and a majority of the time lost as reduced performance while at work. Stewart et al. estimated that employees with depression cost employers $44 billion annually because of healthrelated lost productive time, $31 billion in excess of those without depression. These costs do not include labor costs associated with short- and long-term disability.

Service-Related Issues Specific to Depression and Adults
Rigorous research suggests the following at the level of the health care delivery system:

• Policies, procedures, and physician and staff education to promote the screening, differentiation of anxiety disorders, and minor depression from major depression, as well as to promote optimal use of depression-related medications and mental health staff resources
• Tracking of members being treated for major depression (per HEDIS guidelines) to promote treatment of adequate duration (6 months) and consistency
• Separate tracking of patterns of health care utilization of members with both a depressive disorder and a major medical or behavioral comorbidity
• Resources within every health care delivery system to assure that all adults with likely depressive disorders can be appropriately diagnosed and treated
• Direct outreach by telephone to patients with depression can be of significant value in assuring adherence to prescribed regimens of care and in identifying additional issues to be addressed by medical and ancillary staff.
• Since behavioral disorders—with anxiety and depression most prominent among them—have a major impact on worker productivity, managed care plans marketing their services to employers may wish to consider offering an expanded package of screening and treatment services to reduce worker absenteeism and otherwise improve employee productivity.

Rigorous research suggests the following at the clinic visit:

• Routine screening of all adults for depressive disorders, using two simple questions (“Over the past 2 weeks, have you felt down, depressed, or hopeless?” and “Over the past 2 weeks, have you felt little interest or pleasure in doing things?”) (USPSTF, 2002b, 2003) should be done at most, if not all outpatient visits, with follow-up as appropriate relative to psychotherapy and medication. This screening may be conducted at every primary care visit for otherwise well adults and at every primary care and specialist visit for members with excessive ambulatory care utilization and/or major medical or psychiatric comorbidity.
• Patients selected to receive antidepressant medication for major depression would then be followed for a full 6 months to ensure adherence to prescribed regimens of care and success in addressing depressive symptoms and the complications of depression (per HEDIS guidelines). Whether such patients are managed entirely by the primary care physician or by a mental health professional would depend on the training and comfort level of the primary care physician and the availability of specialized mental health staff and other resources available within the managed care plan or health care delivery system.
• Primary care practitioners may be made aware of the frequency that they are likely to encounter generalized anxiety disorder and depressive disorders in their practice and be made aware of the treatment options within their respective health care delivery systems. This in turn will require the managed care plan or other health care delivery system to develop policies and procedures as well as education and outreach to primary care practitioners and their staff to ensure that the guidelines are understood and effectively implemented. This probably is best done through the use of facilitators reaching out to primary care offices and clinics in the context of quality improvement programming, as described elsewhere in this report.

Evidence for Clinical Benefit: All Adults
According to the 1996 Second Edition of the U.S. Preventive Services Task Force’s Guide to Clinical Preventive Services (USPSTF, 1996)—

It has been repeatedly documented that primary care providers do not recognize major depression in approximately half of their adult patients with this disorder (Schulberg et al., 1985; Borus, Howes, Devins, Rosenberg, & Livingston, 1988; Wells et al., 1989; Coyne, Schwenk, & Smolinski, 1991; Attkisson & Zich, 1990). Because the majority of persons with depression are seen by nonpsychiatrist physicians (Regier et al., 1993), and because effective treatments—drugs, psychotherapy, or a combination of the two—are available for the treatment of depression (Elkin, Shea, Watkings, et al., 1989), it has been proposed that routine depression screening could result in improved recognition and earlier treatment of depression with improved patient outcome (USPSTF, 1996). Clinical trials have shown that use of depression screening tests in primary care settings can increase clinician detection of depression (Attkisson & Zich, 1990; Moore, Lilmperi & Bobula, 1978; Linn & Yager, 1980; Zung, Magill, Moore, & George, 1983; US-PSTF, 1996). Separate research has found that treatment of persons with depression leads to improved outcome (Elkin et al., 1989; USPSTF, 1996).

In a study published in 2001, Schriger et al. (Schriger, Gibbons, Langone, Lee, & Altshuler, 2001) demonstrated a limitation of screening for behavioral disorders. A randomized controlled trial was done in an emergency room setting in which the cases and controls were screened with a 7-minute questionnaire known as PRIME-MD to detect undiagnosed psychiatric illness. In the case group, the physicians were given the report of the screening. In the control group, this information was not provided to the physician. In this study with 92 cases and 98 controls, 42 percent of the patients received a psychiatric diagnosis from the PRIME-MD questionnaire. Only 5 percent of these patients were diagnosed by the physician. Either way, very few of these patients received either additional diagnostic evaluation or treatment for their behavioral disorder—whether diagnosed by the questionnaire or the physician. This study graphically illustrates the need to have policies, procedures, and a system in place if screening for behavioral disorders is to have a favorable impact on behavioral outcomes. Schriger’s conclusion was basically the same as that reached by Schade et al. in a 1998 literature review (Schade, Jones, & Wittlin, 1998) where they found that screening did not necessarily lead to increased medical management of depression.

Tutty et al., in a study of telephone counseling as an adjunct to antidepressant treatment in the primary care system (Tutty, Simon, & Ludman, 2000), documented that a relatively inexpensive telephone outreach system to patients significantly improved depression-related outcomes without affecting the number of visits for treatment of depression. This controlled but nonrandomized study was quickly followed by three more studies which were well-done randomized controlled studies leading to the same conclusion—that enhanced management of depression in primary care settings can significantly improve patient outcomes in a cost-efficient manner.

In a study published in 2001, Katon et al. (2001) used three telephone visits and two visits with a depression specialist. In another randomized trial of telephone support, Hunkeler et al., working in a managed care setting (Hunkeler et al., 2000), demonstrated substantial improvements in depressionrelated symptoms with an intensive nurse telehealth intervention. The intervention consisted of 12–14, 10-minute phone calls from the nurse to the patient over a 16-week period with benefits continuing the duration of the 6-month follow-up period. In a multicenter randomized controlled trial involving 181 primary care practitioners in 46 clinics in six managed care plans, Wells, Schoenbaum, et al., (Wells et al., 2000; Schoenbaum et al., 2001) demonstrated that a quality improvement initiative aimed at improving the quality of the physician and nurse care for depression in these clinics could effectively yield substantial improvements in medication compliance and patient outcomes.

Program Implementation Issues:
Managing Depression Screening and Follow-Up
The prevalence and impact of depression have been demonstrated clearly in both primary care and specialty settings, and the benefits of psychotherapy, cognitive therapy, and pharmacological management likewise have been amply demonstrated in well-done studies. A limited number of well-done studies demonstrate a dramatic costeffectiveness for detection and management of depression in selected patients with one or more major chronic diseases (Vickery et al., 1983; Olfson et al., 1999; Koproski, Pretto, & Poretsky, 1997). Unfortunately, the broader literature is not consistent in findings or quality of study, and many common clinical situations are not addressed. There also are cautionary notes to be considered when addressing depression in patients who have selected chronic diseases, relative to interactions with other drugs being prescribed, and direct adverse effects of selected antidepressive medications on the underlying chronic illness (Wamboldt, Yancey, & Roesler, 1997; Greenberg, Scharf, & Green, 1993; Storch, 1996; Gill & Hatcher, 2000; Goodnick, 2001). Yet another factor is that some patients with one or more major chronic diseases will not be willing to accept either psychotherapy or medication to address their depression (Yohannes, Connolly, & Baldwin, 2001). Fortunately, the available literature suggests that all or almost all patients of all ages and conditions are willing to accept psychoeducational counseling or group sessions to improve their coping skills, stress management, and other behavioral capabilities (Thomas & Weiss, 2000; Spiegel, 1995; Arean, Alvidrez, Barrera, Robinson, & Hicks, 2002). Furthermore, the limited literature in this arena also suggests that group psychoeducational sessions generally are as effective as one-on-one sessions, where group sessions are feasible. The limited benefits available from the psychoeducational interventions may be enough to meet the needs of many of the patients suffering from anxiety disorders and minor depression and may be of limited value to some with major depression or bipolar disorder. For the rest, however, more definitive management of the depression, probably including pharmacotherapy, will be required if optimal outcomes and reduction of other health care costs are to be secured.

In 1997, Lustman et al. (Lustman, Griffith, Freedland, & Clouse, 1997) reported on 5- year follow-up of 25 persons with diabetes who had participated in an 8-week trial of depression treatment. In this follow-up, response to antidepressant therapy was rapid and dramatic, but depression frequently recurred, with 23 (92 percent) of the patients experiencing an average of 4.8 depression episodes over the 5-year period. Presence and severity of depression at follow-up correlated with worse glycemic control and neuropathy. Research supports the approach of frequently rescreening patients who have been treated for depression to detect possible relapse.

In a series of randomized controlled trials of antidepressant treatment of persons with diabetes, published from 1995 to 2000, Lustman et al. demonstrated improvement in both depressive symptoms and glycemic control with nortriptyline (Lustman et al., 1997), fluoxetine (Lustman, Freedland, Griffith, & Clouse, 2000), alprazolam (Lustman et al., 1995), and cognitive behavioral therapy (Lustman, Freedland, Griffith, & Clouse, 1998; Lustman, Griffith, Freedland, Kissel, & Clouse, 1998).

When comanaging depression and a major chronic disease in a given patient, one can expect improvement in both disorders, but in most cases, clinical outcomes are not as favorable for either condition than when dealing with a depressed patient without a chronic disease, or a chronic disease patient without depression. This appears to be because of the interaction of the two sets of disorders and in some cases, interactions between the drugs used to manage the two disorders. The research suggests optimal management of depression, from both preventive and therapeutic perspectives, in patients of all ages, with and without medical comorbidities, involves a multistep process, as follows:

• Detection of symptoms suggestive of depression
• Confirmation of diagnosis and determination as to whether the patient has minor depression, major depression, depression related to bipolar (manic/depressive) disorder, other mental illness, or a purely situational reaction without mental illness. Treatments differ substantially, depending on diagnosis. The possibility exists that treatment of depression related to bipolar disorder as if it were unipolar depression could make things worse. The USPSTF, in its 2002 recommendation for universal screening of adults, makes the following point: “Clinical practices that screen for depression should have systems in place to ensure that positive screening results are followed by accurate diagnosis, effective treatment, and careful follow-up. Benefits from such screening are unlikely to be realized unless such systems are functioning well.” (USPSTF, 2003)
• Decision as to course of treatment (drugs, cognitive behavioral therapy and/or psychotherapy), duration of treatment, and whether or not a psychiatrist or other mental health professional will be involved
• Follow-up to assure 6 months of drug treatment for major depression
• Specialists dealing with specific chronic diseases may be encouraged and enabled to include psychoeducational elements in the education they provide patients for self-management of their chronic disease. These elements may include general coping skills and management of stress, anxiety, and depression.
• These same specialists, according to the literature, also should become expert in the interactions between the various antidepressant medications, the chronic disease they specialize in, and the medications used to manage that chronic disease.
• These educational interventions with psychoeducational components could be made readily available to family practitioners managing such patients without specialist referral.
• Consultation relative to appropriate selection of antidepressant medication also could be made readily available to family practitioners in their management of patients with medical comorbidities.
• Both health care system policy development and extensive physician and nurse education are in order relative to depression for the following reasons:
  • The high prevalence of depression in primary care populations, with an even higher prevalence among patients with major illnesses
  • The wide range of therapeutic options
  • The need for a full 6 months of pharmacotherapy for major depression
  • The reluctance of many patients to be referred to psychiatrists or other behavioral health specialists
  • The relative shortage of psychiatrists and other behavioral health specialists in most health care systems
  • The potential harm of managing a bipolar depressive patient as if he or she were a unipolar depressive patient
  • The 2002 recommendation by the U.S. Preventive Services Task Force that all adult primary care patients be screened for depression, but only if the health care system has the capacity to confirm the diagnosis and follow up as appropriate (USPSTF, 2003; Pignone et al., 2002)
• The need for both primary care and specialist physicians to be familiar with the diagnosis and management of depression in their patients with major chronic diseases
• A continuing high volume of new research and new policy recommendations relating to diagnosis and management of depression
• Circumstances currently surrounding the diagnosis and management of depression are such that annual review of the policies and annual reeducation of the medical staff may be in order, at least over the next few years.

Data To Be Gathered
Refer to Appendix D. Data needs specific to adults and depression are as follows:

• HEDIS parameters relative to outpatient visits and duration of pharmacotherapy relative to outpatient visits and duration of pharmacotherapy for patients diagnosed with major depression
• Incidence and prevalence of major depression and other depressive disorders as determined by claims data, pharmacy data, and/or record review
• Separate tracking of these data for patients with diabetes, asthma, and other major chronic diseases
• Separate tracking should be considered to identify (from claims data) members who appear to be exceptionally high users of outpatient and/or inpatient services for the purpose of flagging members who might benefit from supplemental psychoeducation and/or behavioral health consultation.

Depression in Patients With a Major Chronic Medical or Psychiatric Illness
According to a review of the literature on depression and chronic illness by Katon in 1998 (Katon, 1998)—

Depression can impact chronic medical illness in a number of ways, all of which can unfavorably impact health care costs. In an elderly cohort of 1,711 ambulatory internal medicine patients with a mean of four chronic medical diagnoses, Calahan et al. (Callahan, Hui, Nienafer, et al., 1994) found that patients with depression had mean total outpatient charges of $1,210 over a 9-month period compared with $752 in nondepressed controls. Unutzer et al. (Unutzer, Patrick, Simon, et al., 1997) in an elderly cohort of 2,558 patients from an HMO with a mean of 1.25 chronic medical conditions found that patients with depression had total medical costs over a 1-year period of $1,510, compared with $1,129 in nondepressed controls after adjustment for chronic medical illness. Simon et al. (Simon, Von Korff, & Barlow, 1995), in a similar study, showed health care costs of $4,246 for depressed patients versus $2,880 for nondepressed patients after adjustment for chronic medical illness. These differences were seen at every level of increasing medical comorbidity (Callahan et al., 1994; Unutzer et al., 1997; Simon et al., 1995).

The first of the major ways that depression can affect major chronic illness is through amplification of symptoms. This means that patients with depression can have more symptoms, more severe symptoms, and more functional impairment from those symptoms than nondepressed controls with similar severity of chronic illness. This has been demonstrated by Walker et al. (Walker, Gelfand, Gelfand, et al., 1996) in patients with inflammatory bowel disease, by Fann et al. (Fann, Katon, Uomoto, et al., 1995) in patients with head injury; by Dwight et al. (Dwight, Ciechanowski, Katon, et al., 1997) in patients with Hepatitis C; and by Lustman et al. (Lustman, Clouse, & Carney, 1988) in persons with diabetes. Unfortunately, both primary care physicians and medical specialists can easily confuse worsening of symptoms due to worsening of depression with worsening of the underlying medical condition, leading to unneeded medical testing and unneeded increases in medication dosages (Katon, 1998; Bridges & Goldberg, 1985). Two randomized double-blind studies have shown that effective treatment of major depression is associated with a significant decrease in physical symptoms of chronic medical illness. Sullivan et al. demonstrated this in patients with chronic tinnitus (Sullivan, Katon, Russo, et al., 1993). Borson et al. demonstrated this for patients with chronic obstructive pulmonary disease (Borson, McDonald, Gayle, et al., 1992).

The second major way that depression can affect patients with major chronic illness is by reducing their social and vocational functionality. In these cases, severity of underlying illness and severity of depression seemed to have additive impact on both perceived severity of symptoms and functional disability (Wells et al., 1989). Three papers have shown that severity of functional disability varies over time with severity of depression (Bruce & Hoff, 1994; Bruce, Seeman, Merrill, et al., 1994; Lebowitz, Pearson, Schneider, et al., 1997). Sullivan et al. (Sullivan, LaCroix, Grothasu, et al., 1997) reported that functional impairment in patients with coronary artery occlusion of 70 percent or more at baseline was more highly correlated with symptoms of depression and anxiety than with the number of coronary arteries occluded over a 1-year period. Rovner et al. (Rovner, Zisselman, & Shmuely, 1996) had similar findings in elderly patients with visual impairment. Depression has also been shown to reduce the effectiveness of rehabilitation in older patients with stroke, Parkinson’s disease, heart disease, fractures, and pulmonary disease (Katz, 1996).

Finally, depression can adversely affect a patient’s ability and willingness to adhere to prescribed regimens of care. In a case series exploring this issue, Lin et al. (Lin et al., 2000) noted that 32–42 percent of patients with depression did not refill their initial antidepressant prescriptions—and that this rate was basically the same among those with and without resolution of depressionrelated symptoms. This finding is similar to that found in the control groups of studies demonstrating the value of supplemental interventions to improve compliance with prescribed regimens of care for depression (Tutty et al., 2000; Katon et al., 2001; Schoenbaum et al., 2001). This has also been demonstrated for management of diabetes (Glasgow, 1991); coronary artery disease (Carney, Freedland, Eisen, et al., 1995); participation in rehabilitation following myocardial infarction (Blumenthal, Williams, Wallace, et al., 1982); and persons urged to quit smoking (Anda, Williamson, Escobedo, et al., 1990).

Stated in other terms, diagnosis and appropriate management of depression in a chronic disease patient can reduce health care costs and provide the following patient benefits (Lustman, Clouse, & Freedland, 1998):

• Relief of depression and anxiety
• Restoration of normal sleep and eating habits
• Improved social, occupational, and physical functionality
• Improved pain tolerance
• Improved coping with symptoms of illness
• Decreased preoccupation with symptoms of illness
• Enhanced sexual functioning
• Improved adherence with prescribed regimens of care

Since depression can severely adversely affect the ability of a chronic disease patient to adequately self-manage his or her diabetes, asthma, or other illness, screening for depression and management of depression is of special importance to this group of comorbid patients. There is a substantial body of literature documenting the efficacy of effective management of depression in patients with major chronic diseases in improving health-related outcomes for the patient, while substantially reducing medical complications and use of emergency room visits and hospitalization. An entire section of this literature review is devoted to management of these comorbid patients. Although the same is probably true for a wide range of other behavioral disorders, the evidence base of well-designed randomized controlled trials is strongest for management of major depression.

When only health care costs are considered, screening and enhanced management of depression in primary care patients without major medical or behavioral comorbidities is highly efficacious and costefficient in terms of the dollar cost of the health benefits secured for the patients (Katon et al., 2001; Wells et al., 2000; Schoenbaum et al., 2001; Tutty et al., 2000). These services are cost-efficient, but do not result in reductions in use of emergency room and inpatient services sufficient to generate a return on investment related to reductions in other health care costs within 12 months of program initiation. By contrast, screening and enhanced management of depression in patients with major medical and behavioral comorbidities generates substantial returns on investment in terms of reductions in other health care costs (Katon, 1998; Callahan et al., 1994; Unutzer et al., 1997; Simon et al., 1995).

Intervention
Research indicates that the most appropriate depression-related intervention for adults with major chronic diseases is the same as that for all adults—screening with two questions at every outpatient visit, with follow-up as appropriate. This separate section for adults with major chronic diseases is presented because of the potential for such screening to result in improved management of the chronic disease along with concomitant reduction in health care costs.

Evidence for Clinical Benefit and Impact on Health Care Cost:
Adults With Major Chronic Illness
Egede et al., in a record-review study published in 2002 (Egede, Zheng, & Simpson, 2002), compared 825 adults with diabetes to 20,688 adults without diabetes using the 1996 Medical Expenditure Panel Study. He found that individuals with diabetes were 2.5 times as likely to suffer depression as individuals without diabetes, and that health care costs per diabetic were approximately double for those with depression, compared with those without depression. These general findings held even after adjustment for age, sex, race/ethnicity, marital status, poverty, and comorbidity. Findings were similar in a study by Jiang et al. published in 2001 (Jiang et al., 2001), after reviewing records of patients with congestive heart failure in a single medical center. Jiang et al. found readmission and mortality rates at both 3 months and 1 year to be approximately double for patients with depression, after adjusting for major clinical risk factors. Very similar findings relative to the impact of depression on a major chronic disease were published by Abramson et al. in 2001 (Abramson, Berger, Krumholz, & Vaccarino, 2001) when doing a record review of the risk of heart failure among older persons with isolated systolic hypertension.

In a case report associated with a literature review, Zeigelstein (2001) noted a very high prevalence of depression in patients following myocardial infarction and observed that depression was associated with noncompliance with physician recommendations and increased mortality. His paper did not explore whether management of the depression could have improved patient outcomes.

Asthma and chronic obstructive pulmonary disease are common lung disorders for which tricyclic antidepressants are problematic because of their effect on pulmonary and cardiovascular function (Wamboldt et al., 1997; Greenberg et al., 1993). A number of randomized and nonrandomized clinical trials of short courses of cognitive behavioral therapy (one to 10 visits) have shown significant benefit for symptoms of depression and anxiety and self-management, but not lung function (Perez, Feldman, & Caballero, 1999; Ringsberg, Lepp, & Finnstrom, 2002); for lung function, but not symptoms of depression and anxiety (Kunik et al., 2001; Eiser, West, Evans, Jeffers, & Quirk, 1997); or both (Grover, Kumaraiah, Prasadrao, & D’souza, 2002; Colland, 1993).

E. Depression in High-Cost Patients Without a Major Chronic Disease

In 1998, Panzarino (1998) explored the direct and indirect costs of nontreatment of depression. In this paper, he noted that depression is underdiagnosed in primary care, and that up to 50 percent or more of patients presenting in primary care settings have no diagnosable medical illnesses. The most common symptoms that could not be traced to a known organic cause were back pain, dyspnea, insomnia, abdominal pain, and numbness (Kroenke & Mangelsdorff, 1989). In addition, studies of overutilizers of medical care by Katon et al. (1990, 1992) and Simon (Simon, GE, 1992) showed a high prevalence of psychiatric illness and 68 percent with a past or current history of depression. These data invite consideration of the possibility that screening for and effective treatment of depression might reduce these physical complaints and visits. Katzelnick et al. published a randomized clinical trial of depression management for high users of ambulatory services (Katzelnick et al., 2000). This paper showed dramatic improvements in both behavioral and physical health domains. A follow-up paper a year later (Simon et al., 2001) confirmed the improvements in health indices and an increase in health care costs. A similar follow-up study by Katon et al. (1992) showed similar results—but in this study, the control patients also showed substantial reductions in health care utilization, suggesting the possibility that contamination of the controls with the case intervention may have masked a possible benefit. High users of ambulatory services also are addressed in a separate section of this report dealing with somatization and hypochondriasis.

In 1995, Simon et al. published another paper on health care costs associated with depressive and anxiety disorders in primary care (Simon, Ormel, Von Korff, & Barlow, 1995). In this case series, the authors noted that patients with anxiety or depressive disorders had baseline costs approximately double those with subthreshold disorders or no anxiety or depression, with these differences reflecting differences in medical (as opposed to psychiatric) costs. He also noted that improvement in depression over 1 year of follow-up did not reduce health care costs. In 1997, Simon and Katze