Chemical Dependency and Drug Testing in the Workplace

Alcohol and other drug abuse is estimated to cost the nation over $160 billion per year in lost productivity, health care costs, and crime. There are constant reminders in the news media of the ever-present effects of alcohol and other drug abuse. Health care professionals witness the effects of alcohol and other drug use on the health of the population every day. The annual cost of therapy for alcohol and other drug abuse is $1.6 billion (Gunby 1987; Hawks & Chiang 1986). Drug abuse has extended into all levels of society, including the workplace. To rid the workplace of drugs, the prevalence of drug abuse needs to be known as well as whether, in fact, such abuse is directly related to workplace outcomes, such as productivity, safety, and product integrity. If drug testing is one method of lowering the incidence of drug abuse in the workplace, will it be cost-effective if drug use is less prevalent than has been estimated, and will drug testing be effective in solving ! this problem either by deterrence or remediation?

Much of what is estimated about the use of drugs in the workplace has been obtained from survey or interview data that have been extrapolated to the general population. This is reasonable to do because a sizable proportion of the general population is working. However, some of these data are biased toward overestimation because of their focus on drug abusers within sample populations. Many surveys have focused on specific populations, such as hospital emergency room patients, hospital clinical laboratory results, drug user hot lines, poison control center phone calls, reports from examiners, drug enforcement agency confiscations, and drugs used by intoxicated drivers (Litovitz, Schmitz & Holm 1988; Caplan et al. 1987; Frank 1987; Bailey 1984). While such reports tend to emphasize the seriousness of the drug abuse problem, they do not actually estimate the prevalence of drug use in the workplace.

The credibility of alcohol and other drug use estimates has recently been called into question (Barnes 1988; Schuster 1987; Kozel & Adams 1986). Past estimates from the National Institute of Mental Health (NIMH) on the number of alcoholics in the united States were in part based on a survey questionarire that did not include questions about alcoholism per se (Barnes 1988). Drinking behavior and

† This article originally appeared in the Western Journal of Medicine, Volume 152, Number 5, and is reprinted with permission. Minor editorial and stylistic changes have been made.

*Department of Laboratory Medicine, University of California Son Francisco School of Medicine; Toxicology Laboratories, Northern California Occupational Health Center, San Francisco Department of Public Health.

**Department of Medicine, University of California San Francisco School of Medicine; Divisions of Occupational Medicine and Toxicology, Northern California Occupational Health Center, San Francisco General Medical Center.

Please address reprint requests to John D. Osterloh. M.D., Toxicology Laboratory, Room 3501, Son Francisco Medical Center, 1001 Potrero Avenue, San Francisco, California 94110.

how it is defined have changed over the years, making comparisons to past estimates less intelligible. Larger surveys by the NIDA National Household Survey on Drug Abuse and High School Seniors Survey, have inherent errors due to self-reporting. Adolescent survey respondents tend to overestimate actual drug use. Nevertheless, it is alarming to note that 36.3% of high school seniors have used marijuana and 10.3% have used cocaine in their lifetimes. Studies conducted recently by NIMH have indicated that in a survey of 20,000 people covering five geographic areas across the United States, 6.8% have alcohol abuse or dependence problems and 2% abuse or are dependent on illicit drugs (Barnes 1988). These data still do not indicate the magnitude of the problem of actual drug abuse in the workplace.

Within the medical workplace, 3% of physician have histories of drug dependence and 10% admit to using psychoactive drug use on a monthly basis (McAuliffe et al. 1986, 1984). In randomly selected tractor-trailer drivers whose blood and urine were tested for drugs during truck weighing in Tennessee (Lund et al. 1988), 15% were positive for cannabinoids, 12% for nonprescription stimulants (e.g., phenylpropanolamine), 5% for amphetamine, and 2% for cocaine. Alcohol was detected in only 1% of the sample. In a sample of recruits for the Chicago Police Department, 20% tested positive for cannabinoids (Ostrov & Cavanaugh 1987). Actual drug-testing from various

Laboratories engaged in providing services to industry have indicated that the overall number of positives in preemployment testing is approximately 12% (primarily marijuana), whereas limited random testing data suggest that positive urine tests from currently employed workers is about 2%, primarily due to marijuana. One large testing laboratory has reported the results of testing from seven transportation companies over several years (Whitney 1989). The rate of drug-positive urine specimens varied from 3% to 16% among these companies. These rates reflect a mixture of test settings, including preemployment, random, and probable-cause testing. The data of actual drug use in the workplace are limited. There has been some criticism that the focus on other drugs has been misguided because alcohol abuse is a much larger problem and accounts for a greater portion of morbidity in terms of health effects and accidents (McBay & Hudson 1987).

Even more unclear is the relationship between drug use or abuse and productivity and safety in the workplace. Clearly, workers who are guardians of the well-being of others should not risk the safety of others by using drugs. The general consensus that any psychoactive substance may affect judgment or performance is a safe and cautious one. Drugs are often implicated circumstantially in accidents where a cause is not proved. For example, the often cited air crash aboard the aircraft carrier Nimitz in 1981 did involve drug users. At that time, it was estimated that 47% (both by survey and direct testing) of sailors were positive for at least one illicit drug, usually marijuana (Needleman & Romberg 1989; Hilderbrand 1986; Booz, Allen & Hamilton, Inc. 1981). When this disaster occurred, those killed on board the flight deck were tested and, not surprisingly, half of these sailors tested positive. In this case, the prevalence of drug use by the sailors on board the flig! ht deck merely reflected the prevalence of drug use within the Navy itself and may not necessarily prove a causal role in this accident. As a corollary, the 65 deaths and six accidents aboard naval vessels in 1989 seem to have little relationship to the current 3% prevalence of drug-positive urines tests.

The effects of drugs on human performance are not uniform. While alcohol will clearly alter cognitive and motor function, other drugs are not so well studied. Even in the forensic area of drug-intoxicated drivers who are arrested following accidents or because of the appearance of driving while intoxicated, experts do not agree on which types of human function might be altered by individual drugs (American Medical Association Consensus Development Panel 1985). For example, marijuana use may not affect reaction time in terms of stimulus-response reflex, but it can reduce attentiveness (McBay 1988; Jones 1987; Mason & McBay 1985). Generally the predominant effects of marijuana on behavior and physiology do not last longer than four hours after smoking a marijuana cigarette, but decrements in performance during complex situations among airline pilots have been measured as long as 24 hours after a dose (Yesavage et al. 1985). In conflict with the general assumption that mar! ijuana affects workers' performance are the findings of a study that allowed workers to buy either tobacco cigarettes or marijuana cigarettes. Those using their money to acquire marijuana and smoking as many as 12 marijuana cigarettes per day for the three-week period of the study did not show a difference in work output (Mello & Mendelson 1985).

Drug use and abuse is a real problem in society, but the effects of drug use on worker performance and safety need better documentation in order to accurately address these problems. The study of actual illicit drug use, defined dosages, temporal relationships, and effects on work-modeled or real workplace tasks should be addressed with respect to each specific drug in question.

Workplace drug testing may connote both punitive and remedial intent. The anticipated effect of workplace drug testing would be to deter drug use if the consequences of detection are undesirable to the individual user. Detection alone is of little value unless the drug use is discontinued. Most major corporations that have embarked

on drug-testing programs have taken remedial action rather than punitive action by providing employee assistance programs (EAPs). Such programs may reduce costs by providing rehabilitation of an existing, well-trained employee, instead of having to train a new worker. The overall goals of employee drug testing include improvements in workplace safety, productivity, and product integrity so that the individual industry will have an improved fiscal bottom line.

Approaches to drug testing have included preemployment, probable-cause, safety-sensitive, and random testing. Preemployment testing is the most popular type of drug-testing program and perhaps the most effective with the least liability to the employer. Between 60% and 70% of small businesses in one eastern city (Godefroi & McCunney 1988) and 80% of Fortune 500 companies with drug-testing programs use the preemployment testing approach (Noel Dunivant & Associates 1985). Over a period of years, preemployment drug testing would be expected to lead to a lower incidence of drug use in the work population when drug users are screened out prior to being hired and old workers leave the work force by natural attrition. Probable-cause testing is useful in enhancing the presumption of drug use when poor performance is observed, evidence of intoxication is witnessed, or after accidents thought to be caused by intoxication. Supervisors and employees can be instructed to recogni! ze the signs of drug intoxication. While the detection of drugs in urine has no accurate relationship with behavior or pharmacologic effects, combining witnessed behavior with drug detection can provide a reasonable presumption that drug abuse or drug effects were present.

The random testing of workers by industry is a less popular approach in that it has several shortcomings. To identify most of the drug users in a population using a random testing approach, urine specimens must be collected frequently compared to the period of time that drugs may be present in the urine (see Table I). Random testing should be unannounced so that drug-using workers may not anticipate the time of drug testing and abstain from drug use during that period. This requires an added level of administration and management that may be more costly. While many approaches to drug testing have been examined in the courts (e.g., safety sensitive, probable cause, and preemployment), the legality of random testing programs has not been decided. Because of the potential for infringing on the right of privacy of individual workers (see legal issues section) some industries are steering away from random testing (Hoerr et al. 1988). While the specificity of drug testing (low fa! lse-positive rates) is good when performed properly, low efficiency of such testing may occur when the prevalence of drug abuse is also low. Whenever the prevalence of a test condition (ie., drugs being present) approaches the specificity of a test, the number of false-positives tests relative to the number of true-positives may be quite high. For instance, phencyclidine (PCP) is a relatively rare drug encountered in workplace drug-testing programs. If one in 1,000 workers actually had PCP in their wine and only one in 1,000 false-positives occurred in the testing procedure, then half of the positive results in such testing would be false-positive. Clearly, random testing should not be applied to types of drugs with very low prevalence.

This raises the question of which drugs should be tested. When working with laboratories, employers have the choice to select tests for drugs they think are problematic. Marijuana and cocaine are the drugs most often detected in surveys and programs. Opioids, amphetamines,

Journal of Psychoactive Drugs ***** Page 409 ***** Vol. 22(4), Oct-Dec 1990

Osterloh & Becker ********** Drug Testing in the Workplace

and phencyclidines are less common. Other drugs, such as barbiturates and, benzodiazepines, may also be included in such testing because their use often is encountered among polydrug users. Some drugs in certain regions of the country or in certain industries may not be perceived as a problem. If that is the case, employers may wish to spend their money in another area with a higher rate of return.

Recently, two studies have shown minimal effects of preemployment drug testing. One study screened all, new employees and kept results confidential for a year (Parish 1988). Of all the new employees, 12% had tested positive for drug use. There were no differences between drug-positive and drug-negative employees when job performance variables were compared (e.g., evaluations, job retention) at the end of the year. Eleven drug-negative employees were fired and no drug-positive employees were fired during the period of this study. The other study evaluated 3,600 workers who were hired by the U.S. Postal Service between 1987 and 1988 (Unsigned 1989). Overall, 8.4% of all workers hired had positive drug tests; 13.3% of workers with positive drug tests were fired within six months compared to 9.5% of workers with negative drug tests. While the effects of drug use on work performance have been difficult to document, uncontrolled data from industry have suggested that drug testing! in the workplace has been efficacious. Southern Pacific Railroad has indicated that its accident rate has decreased from 10% to 5% over a three-year period of drug testing (Taggart 1989). General Motors has claimed a greater than 40% reduction in absenteeism, 50% fewer disciplinary actions, and 50% fewer accident claims by employees (Castro et al. 1986). The drug-testing program of the United States military has been exemplary. In the Navy, the most common drug detected was marijuana. The overall positivity rate has fallen from 48% in 1980 to 3% in 1987 (Needleman & Romberg 1989; Booz Allen & Associates 1981). While these numbers reflect actual test positivity rates, the numbers from 1980 must be viewed with some skepticism because of the different testing techniques and criteria used in 1980 versus 1983 and thereafter (Needelman & Romberg 1989; Hilderbrand 1986).

Because observations in industry have been uncontrolled, it may be argued that drug use is reduced by deterrence and fear of consequences, but that improvements in safety or productivity are coincidental. For instance, when accidents occur, attentiveness is subsequently heightened and preventive measures are taken that may reduce accident rates. Also, drug testing has been instituted at a time of labor union weakness and union-busting tactics. Employees may refrain from drug use for fear of losing their jobs in this general milieu. The Hawthorne effect can be postulated to be at work because this extra attention given to workers and their conditions may improve productivity and morale. Cause and effect has not been documented. Even a corollary hypothesis may be invoked: poor working conditions with attendant unsafe practices, low worker morale, and low productivity may lead to increased drug use. Overall, drug testing has been viewed as successful in highly regimented situa! tions, such as the military, or in highly focused applications, such as preemployment or probable-cause testing.

The legal issues that have surfaced in the development of drug-testing programs have been reviewed extensively (Chamberlain 1989, 1988; Long 1989; Substance Abuse Testing Committee 1988; American Medical Association Council on Scientific Affairs 1987a; Imwinkelried 1987; Adams & Remmers 1986; Angarola 1984). Concerns have focused primarily on the right to privacy or lack of probable cause as related to the Fourth Amendment, equal protection with respect to the application of law under the Fourteenth Amendment, the application of due process and the admissibility of drug test results under the Fifth Amendment, and contract law with respect to policy agreements between employers and employees. Also of consideration are government legislation and local law.

While several constitutional and human rights issues have been evolving during the implementation of drug testing in the 1980s, the legal cornerstone of drug-testing programs is the emplacement of a policy agreement. Such a policy is a contract between employers and employees. As such, it should be negotiated in advance by both parties and agreed on with respect to the conditions of testing, the procedures for collecting specimens, and the consequences of the test results. Each step in this pathway should be definitively described so that choices and contingencies are made by the policy itself and not by others acting after the fact. Employers should guard that policies are evenly applied and nondiscriminatory, and allow the tested employee adequate recourse to any decision made regarding the employee. Furthermore, each program should include an EAP. In adopting such policies for drug testing, employers should be aware that local regulations or state legislation may have re! cently been enacted that affect how such programs would be conducted.

Selecting a group of persons to test without prior suspicion may lead to an allegation of bias or an absence of equal protection. In a ruling against the National Collegiate Athletic Association, the Fourteenth Amendment was cited, in part, as protection against the selecting a group of athletes for drug testing without demonstrating a likelihood that drug use was prevalent in that population. In this particular circumstance, college athletes other than male football and basketball players were found not to

Journal of Psychoactive Drugs ***** Page 410 ***** Vol. 22(4), Oct-Dec 1990

Osterloh & Becker ********** Drug Testing in the Workplace

have a prevailing drug problem that required testing based on the actual test results in each sport. Due process must be given to a person considered for drug testing. This applies not only to the acquisition of a urine specimen as described by policy procedures, but also with respect to the admissibility of drug-testing evidence in court actions. Drug-testing techniques must have general scientific acceptance, but they are not required to have absolute accuracy, only beyond reasonable doubt.²

The most difficult area of constitutional law relates to the Fourth Amendment, which prohibits the federal government from unreasonable search-and-seizure activities. In most cases where there was probable cause, drug-testing procedures have been upheld. When applied randomly (i.e., without individualized suspicion), the courts have had difficult decisions before them. Some legal opinion has extended the prohibition of unreasonable search and seizure to an intrusion into individual private lives for both government and private sectors. Other interpretations of the Constitution have suggested that the authors of the Constitution did not intend to protect the private act of illicit drug use, but were concerned with the privacy of family life, public service, and economic practices. Executive Order 12564 by President Reagan in 1986 has been applied slowly, primarily to the random testing of employees in safety-sensitive positions, wherein the public concern for safety (e.g., a! irline pilots) or corruption of a process (e.g., customs agents) has been held as a greater concern for the majority than the rights to privacy of the individual. For example, in prison guards who were searched and tested for drugs, a lower court ruled that probable cause should be required before testing. A higher court reversed this decision, indicating that lack of drug use was necessary for the job function being performed.³ Drug testing of employees in certain critical positions has been favored as being a type of fitness test.

Several cases have been decided in favor of the plaintiff because of invasion of privacy and protection under constitutional law.^4,5 Rulings in favor of the plaintiffs in these cases were contingent on the lack of probable cause and the poor implementation and practice of a policy agreement. A higher court ruling may still be heard these cases. Also, there are examples of several local jurisdictions that have tried to coerce employees into providing urine for drug testing without having adequate policy agreements. Such actions, no matter how likely the actual drug problem, were doomed to fail.

For years, many laboratories have been capable of measuring drugs and their metabolites in biological fluids. These measurements have been used in a number of different circumstances, including clinical emergency toxicology (drug overdose cases), medical examiners' investigations (cause of death determination), methadone-testing programs and therapeutic drug-monitoring practices. Until recently, the many different techniques and approaches for the detection of drug use have been unique or suited to those particular situations, but widely varied between them. With the development of the NIDA guidelines for federal employee drug-testing programs in 1998, there has been general consensus regarding the types of methods to be used in such testing (American Medical Association Council on Scientific Affairs 1987b; Hoyt et al. 1987).The following conditions cause employee drug testing to be different from other forms of drug testing:

Testing for marijuana and cocaine, two of the most commonly abused drugs, appears to be most productive, but the advocates of drug-testing programs have been criticized for not including the most common of all abused drugs --- alcohol. As stated earlier, programs that focus on drugs abused in very low prevalence may also be misguided because false-positives may outnumber true-positives when the prevalence approaches the specificity of the testing. NIDA (1988) has prescribed that the initial screening test be an immunoassay (radioimmunoassay, enzyme immunoassay, fluorescent polarization immunoassay) for the five major illicit drugs of abuse: marijuana, cocaine, amphetamines, PCP, and opioids. Positive results

Journal of Psychoactive Drugs ***** Page 411 ***** Vol. 22(4), Oct-Dec 1990

Osterloh & Becker ********** Drug Testing in the Workplace

TABLE II NIDA-RECOMMENDED CUTOFF CONCENTRATIONS IN URINE FOR WORKPLACE DRUG TESTING
	Immunoassay Screen	GC-MS Confirmation
Drug	(ng/ml)	(ng/ml)
Marijuana metabolites^a	100	15
Cocaine metabolite^b	300	150
Opioid metabolite	300^c	300^d
Phencyclidine	25	25
Amphetamine	1000^e	500^f
^{adelta9-tetrahydrocannabinol-9-carboxylic acid. ^{bBenzoylecgonine. ^{cTotal conjugated opiates; 25 ng/ml if by radioimmunoassay for free morphine. ^{dEither codeine or morphine after hydrolysis. ^{eTotal reacting amphetamines.}}}}} fEither d-amphetamine or methamphetamine.

by the initial screening methods would then be confirmed subsequently by gas chromatography-mass spectrometry (GC-MS). The cutoff (detection of the presence of the drug) concentrations of these methods are listed in Table

No method has absolute accuracy, and any can be fooled by similar drugs or intentional adulteration. Drug-testing results may be inaccurate because of analytical false-positive or false-negative findings or owing to the misapplication of the results. Usually, drugs or chemicals that produce false-positives or false-negatives in the screening assay will not produce the same problem in the confirmatory assay. However, agents initially causing false-negatives might not produce further confirmatory testing. Therefore, the aim of drug users wishing to avoid detection has been to interfere with the initial immunoassay screening test. Fluorescent polarization immunoassays for barbiturates and benzodiazepines can be rendered falsely positive by the use of nonsteroidal anti-inflammatory agents, such as ibuprofen (Larsen & Fogerson 1988). Also, the presence of fluorescein dye in the wine may prevent determination by fluorescent polarization immunoassays (Elin et al. 1989; Inloes,! Clark & Drobnies 1987). GC-MS procedures can also be fooled. It has recently been shown that ibuprofen will cause a false-negative result in some GC-MS confirmatory tests for marijuana metabolites (Brunk 1988). This is probably due to the consumption of the derivatizing agent by ibuprofen, which is necessary to make marijuana volatile for gas chromatographic analysis. Other immunoassays, such as enzyme immunoassays, have been shown to produce false-negative results by additives, such as benzalkonium chloride (an antiseptic used topically and as an antimicrobial preservative in ophthalmic solutions, such as Visine eye drops), vinegar, lemon juice, golden seal (Hydrastis canadensis) tea, lye crystals for unplugging drain pipes, and liquid soap. All but the benzalkonium chloride are required in high concentrations, and are easily detected by simple means, such as determination of pH, specific gravity, shake tests for foaming, creatinine concentrations, and osmometry. The use! of benzalkonium chloride from Visine eye drops has been shown to cause false-negative results in enzyme immunoassays for marijuana and cocaine, and can only be ruled out by testing specifically for benzalkonium chloride (Hurrell & Spiehler 1989; Pearson, Ash & Urry 1989; Warner 1989; Mikkelsen & Ash 1988). A recidivistic drug user may use other approaches, such as diluting the urine specimen with toilet water, substituting the urine sample with a known drug-free urine specimen, ingesting substances that dilute the urinary concentration of the drug to be detected (e.g., drinking water, taking diuretics), ingesting acidic salts (e.g., ammonium chloride) to hasten the excretion of basic drugs before the urine collection or ingesting basic salts (e.g., sodium bicarbonate) to reduce the excretion of basic drugs during collection.

Apart from these purposely added or ingested adulterants that may cause false results, a potential problem is the misidentification of drugs taken innocently without pharmacologic effects or taken therapeutically. Usually, the confirmatory testing procedures can rectify the misidentification caused by a cross-reactivity of the immunoassays with similarly structured drugs. Drugs that cross-react with the initial screening immunoassay test are listed in Table III. For the opioids, medications with codeine, hydrocodone (e.g., certain cough syrups) and oxycodone will cause the initial immunoassay to be positive

TABLE III

COMMON CROSS-REACTING DRUGS

IN SCREENING IMMUNOASSAYS

Amphetamines*

Phenylpropanolamine

Pstudoephedrine

Ephedrine

Phenylephrine

Phentermine

Propylhexedrine

l-Methamphetamine**

Fenfluramine

Other amines

Opioids

Oxycodone

Hydrocodone

Hydromorphone

Codeine

Marijuana Metabolite

NSAIDs*** before 1986;

Naltrexone, nabilone

do not cross-react

Cocaine Metabolite

Local anesthetics and atropine

do not cross-react

*Several manufacturers (e.g., SYVA, Abbott) now make a more specific immunoassay for amphetamines that cross-reacts less with other amines.

**In Vicks Inhaler (as desoxyephedrine).

***NSAIDS=nonsteroidial anti-inflammatory drugs.

for opioids. Subsequent GC-MS will differentiate these drugs from morphine. Similarly, over-the-counter stimulants, such as dietary aids and decongestants (e.g., phenylpropanolamine, fenflurarnine, ephedrine) may cause the initial immunoassay screen for amphetamines to be positive. Again, GC-MS differentiates these ffm illicit amphetamines. Consuming poppy-seed-containing foods (e.g., bagels, poppy-seed cakes) will produce detectable amounts of morphine in the urine without sufficient amounts reaching the systemic circulation to produce any pharmacologic effects. Usually the morphine derived from the use of heroin (diacetylmorphine) must be metabolized to monoacetylmorphine and morphine. Therefore, to differentiate poppy seed use from heroin use, the presence of monoacetylmorphine in a GC-MS analysis will implicate heroin use (ElSohly et al. 1988; Mulé & Casella 1988).

Actual false-positive tests are rare when these two testing procedures are combined. The only standardized way of assessing the accuracy of employee drug testing is by providing proficiency test samples for laboratories to test in an open or a blind fashion. Recent studies on employee drug testing have shown that laboratories using these methods are highly accurate, having very low false-positive rates when focused on only the five drug classes (Frings, Battaglia & White 1989; Davis, Hawks & Blanke 1988). A number of authors have concentrated on proficiency testing results that are not applicable to employee drug testing. Unfortunately, these results have also received media attention. Most of the previously published reports on proficiency testing results have surveyed laboratories doing work for methadone drug-testing centers and hospital emergency toxicology services. As previously, the concentrations of drugs, the number and types of drugs, the prevalence of dru! gs in the population tested, and the techniques used in these different situations are not comparable.

Calibration and in-house quality-control checks are routine in any laboratory. In addition, the NIDA guidelines for contractors with federal agencies require that 50% of all specimens submitted by the agency in the first quarter should be blind quality-control specimens. About 80% should be negative urines. In subsequent quarters, 10% of the specimen load is to be for the purposes of blind quality control. It is required that the cost for preparing and testing of these samples be borne by the submitting agency. Also, 10 proficiency testing samples are sent every two months to the laboratories certified by the NIDA program. Any false-positive test requires consideration for recertification of the laboratory and a reanalysis of previous contract specimens. To date, of the initial 100 that applied, only about 50 have been certified by NIDA. More than 50 had dropped out because of the stringent requirements, including a one-week inspection and operations observation. Others hav! e had difficulty because of the added requirement of quantitation (determine concentration) during the GC-MS confirmatory procedures.

The reliability of any test depends on the prior probability of the test condition being present and the analytical sensitivity and specificity of the test. If the tests for marijuana and PCP are both 99.9% specific (0.1% false-positives) and these two tests are applied to a population of 1,000 persons where only one person has PCP in their urine and 100 have marijuana metabolite in their urine, the reliability (predictive value, posterior probability) of a positive test for marijuana in this situation is 99% and only 50% for PCP. While both tests have the same analytical specificity, the relative number of false-positive to true-positive tests will be much higher in the low-prevalence situations (i.e., PCP).

In addition to urine testing to detect drug use, hair analysis has been investigated. Many drugs have been detected in hair, but the interpretations of these tests are vague at present. Hair samples can be obtained repeatedly and noninvasively, can be identified as belonging to a

Journal of Psychoactive Drugs ***** Page 413 ***** Vol. 22(4), Oct-Dec 1990

Osterloh & Becker ********** Drug Testing in the Workplace

particular person, and am less likely to be adulterated; however, drugs present in hair cannot substantiate current use. Therefore, hair analysis is unlikely to be used in probable-cause testing. Also, as a specimen, hair suffers potentially from external passive contamination. Hair analysis was reviewed recently by Baumgartner, Hill and Blahd (1989).

Before establishing a drug-testing program and developing the necessary policy to make such a work, the need for such a program should be established and documented. If a drug abuse problem has been shown to be present in a particular workplace, it is more likely that all those concerned --- employers, employees, unions, regulatory agencies, and legislative or legal bodies --- will be able to join forces in an effort to help solve the drug problem. In the absence of a clear need for a drug-testing program, accusations of bias or discrimination may be charged more easily. Documenting a drug abuse problem may take several forms, including witnessed behavior of intoxication or drug dealing on the job, increased accidents in an environment of suspected drug use, or anonymous pilot testing of employees to document the presence of drugs. In one workplace where almost 500 preemployment tests were done --- at a cost of over $16,000 for the testing alone --- only one PCP-positive e! mployee was found (Bompey 1986). This might not be considered cost-effective, but if that individual was to cause a costly accident or error, these dollars may have been well spent.

The development of a policy for drug testing should include the following (NIDA 1988, 1986; Substance Abuse Testing Committee 1988):

Following agreement among the participating parties, the actual procedure should be defined. If random testing is to be done, choosing the appropriate time for collection may present some difficulty. If employees are aware or notified that collections will occur, the purpose of the program may be defeated. If random collections are not frequent enough, testing would tend to identify only continuous abuse. Random collections may be made at times when other activities are scheduled within the operations of an industry. For example, collections may be made during routine medical examinations (annual physicals), corporate health surveys, operations shutdowns for preventive maintenance or safety inspections, and insurance enrollment periods. Other random collection techniques include the checking of random portions of the worker population. For instance, using random number generation, only 10% of the work force would be sampled.

Other specimen collection schemes relate to the program, whether probable cause, safety sensitive or preemployment. Probable-cause testing is the most common program for testing currently employed workers. The written documentation of witnessed behavior that corroborates test results is extremely important in such a program. Employers should be instructed as to what signs to observe in a worker who is suspected of intoxication. Because probable-cause testing only tests individuals who have been presumed to use drugs because of their behavior or symptoms, only the worst offenders are likely to be identified. It is usually preferable that a worker's supervisor document his or her observations and accompany the individual who is under suspicion of drug use to the location where a urine specimen would be collected.

The urine specimen should be collected in an area away from use by other employees so that the occurrence of testing does not become public knowledge. The direct observation of urination is not absolutely necessary; but some precautions should be taken. The person to be tested should not be allowed to carry extra garments, purses or packages into the toilet area. The toilet area should be free from other cups, containers or water faucets. The toilet water should be dyed with a commercial sanitary indicator. Although drug users in methadone programs or the criminally incarcerated may have sophisticated techniques to adulterate urine or escape detection, such sophistication is not commonplace in the work force. Therefore, direct observation is considered more of an invasion privacy than a necessary or valid procedure. Once the urine specimen is collected, the temperature should immediately be measured and recorded; it should be within 2°C of body temperature. The urine s! pecimen should then be closed and sealed. The employee and the person who is witnessing the collection process should initial and date the seal or label of the container and sign a log book that is kept on the premises. Special tamper-proof containers with seals that will tear irreparably when opened can be purchased for such drug testing. Specimens to be transported

Journal of Psychoactive Drugs ***** Page 414 ***** Vol. 22(4), Oct-Dec 1990

Osterloh & Becker ********** Drug Testing in the Workplace

to laboratories for testing should be packaged in tamper-proof courier bags or locked boxes with the keys or combinations known only by the laboratory. Attached either to an individual specimen or the box of specimens should be a list of individuals who have handled the specimens and their signatures should record their contact with the specimens (NIDA 1988,1986).

The laboratory should report only confirmed positive test results; all other test results are to be reported as negative. Reporting results such as "trace" or "'positive but unconfirmed" or ""unknown substances present" should be strongly discouraged. Policy should provide for retesting of the sample at a laboratory of the employee's choice. Medical review officers (MROs) are considered an important component of drug-testing programs and are mandated for federal programs (NIDA 1988). Their duties include understanding how the tests were performed, how specimens were acquired, how to interpret the results, and what courses of action can be taken. A designated MRO also serves as a person outside the corporate structure who can confidentially review the results of testing and the circumstances leading to positive test results with the employee. In so doing, the MRO must understand what are the cutoff concentrations for individual drugs and the period of time during which indiv! idual drugs can be detected. The MRO must also understand which drugs and conditions may alter the results of drug testing. Most important, the MRO must be able to communicate the meaning of test results to administrative personnel who may take action against the individual employee.

The courses of action available to an MRO should be guided explicitly by policy. The aIternatives, that the MRO may choose from include no action, explanation by the employee of conditions leading to the positive test, an investigation of medications that may have been prescribed, repeating the test with a full explanation of consequences when a second specimen is positive, a change of position within the company, a leave of absence, further performance evaluation, resignation, and termination. With all of the above alternatives, counseling and treatment should be provided through a practitioner with experience in drug abuse treatment and rehabilitation.

Chemical dependency in the workplace may present as acute intoxication, chronic abuse, withdrawal, or as accelerated medical complications of chemical dependency, such as liver disease. Intoxication, withdrawal, and medical complications of chemical dependency are more obvious but occur less frequently than the more subtle, less specific sips of drug use, such as absenteeism, tardiness, poor productivity or accidents.

The key to intervening and treating a drug abuser is early identification and institution of care (Austin 1987; Peterson et al. 1987). As an outgrowth of occupational alcoholism, programs, the majority of Fortune 500 companies now have some form of EAR Employers can encourage employees to use such programs. Because of the tendency for a troubled employee to deny the extent of his or her addiction, intervention techniques have been developed for the purpose of directing identified persons into a structured program. Denial may go beyond the affected person to involve coworkers, supervisors, and family. An effective, intervention makes clear to a chemically dependent worker that both their health and economic livelihood are at stake. The clinicians conducting the intervention may use family members and coworkers familiar with the troubled employee's deteriorating physical and emotional status.

After an appropriate clinical evaluation is made by either a member of the company's EAP or a consulting clinician, the next step is directing the substance-abusing person into treatment (Walsh & Hingson 1985). In some case evaluations, what is perceived as behavioral problems associated with possible drug use may be the clinical manifestations of a primary emotional disorder, such as depression. If chemical dependency is diagnosed, the employee might be referred to a 28-day inpatient drug abuse treatment program. If a primary emotional disorder is diagnosed, a referral may be made for outpatient psychotherapy. Treating an emotional disorder will not be effective unless chemical dependency is also treated; and the 28-day inpatient treatment programs have become the standard for treatment of chemical dependency. The cost of such programs is high, and for some employees with less than optimal health insurance coverage, a more circumscribed period of inpatient care can be ! followed by care in a halfway house or residential treatment setting.

Several models exist for the treatment of chemical dependency (Garvin 1987).The need for self-participation is stressed by Alcoholics Anonymous (AA) or Narcotics Anonymous (NA). For the recovering alcohol or other drug addict, this self-help group approach is often crucial in I a state of sobriety. Most treatment programs use some form of group psychotherapy in addition to the 12-Step approach of AA. It is often only through the group

Journal of Psychoactive Drugs ***** Page 415 ***** Vol. 22(4), Oct-Dec 1990

Osterloh & Becker ********** Drug Testing in the Workplace

context that the chemically dependent worker can come to terms with his or her long-term avoidance of the acknowledgment of the addictive process and its negative effects on the person's life. Other psychotherapeutic modalities may include individual and family therapy.

Chemically dependent workers usually require a defined structure, guidelines, and rules. The inpatient treatment model is often the best means for initially creating such a framework. A medical history and physical examination with appropriate laboratory testing are necessary on entry to treatment. The greatest chance of success with a chemically dependent worker occurs when the program emphasizes total abstinence, but even with such measures, slips do occur. Chemically dependent adults must consider themselves as being in the process of recovering on an indefinite basis. Persons with alcoholism may require pharmacologic support with disulfiram (Antabuse) to decrease their recidivism. A medical evaluation may indicate the need for an antidepressant medication or lithium carbonate for treating an associated affective disorder.

The occupational physician should be a contributor to the design of EAPs (Peterson et al. 1987), which may be situated within the company or outside the company as a contracting service. All EAPs should provide evaluation, treatment referral, follow-up, and employee education. Other services include short-term treatment, management training, drug monitoring, and organizational consultation to departments in the company. A properly run maintains confidentiality in self-referral cases and makes clear the limited confidentiality in a management referral. Other planning functions of the medical consultant include a review of health insurance benefits for treatment of substance abuse and psychiatric illness. Part of the cost-effectiveness of EAPs is the educational effort directed at management and employees about the hazards of substance abuse. Managers can be instructed about the early warning signs of drug abuse. Employees can learn of the acute and long-term effects of alcoh! ol and other drug use. The ultimate goal in such educational efforts is to encourage chemically dependent workers to seek assistance for their addictive disease.

As difficult as these problems are to treat, great satisfaction can be had in helping a dysfunctional, chemically dependent worker to return to a premorbid level of occupational and personal functioning. The professional challenge is great in overcoming the resistance to treatment on the part of the employee and the negative feelings engendered on the part of the health practitioner. While there is a need for flexibility in approaching any troubled person, companies and their consulting health care team should also have a set of policies and procedures for addressing the problems of chemically dependent and emotionally disturbed employees.

Adams, W.F. & Remmers, C.L. 1986. Drugs and alcohol in the workplace: Technology, law and policy. Santa Clara Computer High-Technology Law Journal Vol. 2: 305-388.

American Medical Association Consensus Development Panel. 1985. Drug concentrations and driving impairment. Journal of the American Medical Association Vol. 254: 2618-2621.

American Medical Association Council on Scientific Affairs. 1987a. Issues in employee drug testing. Journal of the American Medical Association Vol. 258: 2089-2096.

American Medical Association Council on Scientific Affairs. 1987b. Scientific issues in drug testing. Journal of the American Medical Association Vol. 257: 3110-3114.

Angarola, R.T. 1984. Drug detection programs in industry. PharmChem Newsletter Vol. 13: 1 -11.

Austin, R.L. 1987. Employee assistance programs. Clinical Chemistry Vol. 33(Suppl.): 101B-105B.

Bailey. D.N. 1984. Comprehensive toxicology screening: The frequency of finding other drugs in addition to ethanol. Clinical Toxicology Vol. 22: 463-471.

Baumgartner, W.A.; Hill, V.A. & Blahd, W.H. 1989. Hair analysis for drugs of abuse. Journal of Forensic Science Vol. 34: 1433-1453.

Bompey, S.H. 1986. Drugs in the workplace: From the batters box to the boardroom. Journal of 0ccupational Medicine Vol. 28: 825-832.

Booz, Allen & Hamilton. Inc. 1981. Urinalysis Test Results Analysis 1980. Arlington, Virginia: Booz, Allen & Hamilton, Inc.

Brunk, S.D. 1988. False negative GC-MS assay for carboxy THC due to ibuprofen interference. Journal of Analytical Toxicology Vol. 12: 290-291.

Caplan, Y.H.; Ottinger, W.E.; Park. J. & Smith, T.D. 1985. Drug and chemical related deaths: Incidence in the state of Maryland 1975-1980. Journal of Forensic Science Vol. 30: 1012-1021.

Castro, L; Beaty, J; Dolan, B. & McDowell, J. 1986. Battling the enemy within. Time March 17: 52-61.

Chamberlain, R.T. 1989. Legal considerations in drug use testing: Privacy rights, contracts and wrongful use of test results. Journal of Forensic Science Vol. 34: 1477-1481.

Chamberlain, R.T. 1988. Legal issues related to drug testing in the clinical laboratory. Clinical Chemistry Vol. 34: 633-636.

Davis, K.H., Hawks, R.L. & Blanke, R.V. 1998. Assessment of laboratory quality in urine drug testing --- a proficiency testing pilot study. Journal of the American Medical Association Vol. 260: 1749-1754.

Elin, R.I.; Bloom. J.N. & Herman, D.C.; Sliva, C.A.; Ruddel, M.; Nussenblatt, R.B. & Palestine, A.G.. 1989. Interference by intravenous fluorescein with laboratory tests. Clinical Chemistry Vol. 35: 1159-1160.

Journal of Psychoactive Drugs ***** Page 416 ***** Vol. 22(4), Oct-Dec 1990

Osterloh & Becker ********** Drug Testing in the Workplace

ElSohly, H.N.; Stanford, D.F.; Jones, A.B; ElSohly, M.A.: Snyder. H. & Pedersen, C. 1988. Gas chromatographic/mass spectrometric analysis of morphine and codeine in human urine of poppy seed caters. Journal of Forensic Science Vol. 33: 347-356.

Frank, R.S. 1987. Drugs of abuse: Data collection systems of DEA and recent trends. Journal of Analytical Toxicology Vol. 11: 237-241.

Frings, C.S.; Battaglia, D.J. & White, R.M. 1989. Status of drugs-of-abuse testing in urine under blind conditions: An AACC study. Clinical Chemistry Vol. 35: 891-894.

Garvin, F.H. 1987. Laboratory assessments in treatments of substance abuse: Utilization and effects on treatment design. Clinical Chemistry Vol. 33(Suppl.): 95B-100B.

Godefroi. R. & McCunney, R.J. 1988. Drug screening practices in mull businesses --- a survey [Letter). Journal of Occupational Medicine Vol. 30: 300,302.

Gunby, P. 1987. Nation's expenditure for alcohol, other drugs, in terms of therapy, prevention now exceed $1.6 billion. Journal of the American Medical Association Vol. 258: 2023.

Hawks, R.L. & Chiang, C.N. (Eds.). 1986. Urine Testing for Drugs of Abuse. NIDA Research Monograph 73. Rockville, Maryland: NIDA

Hilderbrand, R.L. 1986. The U.S. Navy testing program. California Association of Toxicology Newsletter Winter: 22-24.

Hoerr, J.; Hafner, K.M.; Degeorge, G.; Feild, A.R. & Zinn, L. 1988. Privacy. Business Week March 28: 61-68.

Hoyt, D.W.; Finnigan. R.E.; Nee, T.; Shults, T.F. & Butler, T.J. 1987. Drug testing in the workplace --- Are methods legally defensible? A survey of experts, arbitrators and testing laboratories. Journal of the American Medical Association Vol. 258: 504-509.

Hurrell, Y. & Spiehler, V. 1989. Adulterants intended to cause false negatives in urine screens for marijuana or cocaine use. Clinical Chemistry Vol. 35: 1175.

Imwinkelried, E.J. 1987. Some preliminary thoughts on the wisdom of governmental prohibition or regulation of private employee urinalysis testing. Clinical Chemistry Vol. 33(Suppl.): 19B-32B.

Inloes, R.; Clark, D. & Drobnies, A. 1987. Interference of fluorescein, used in retinal angiography, with certain clinical laboratory tests. Clinical Chemistry Vol. 33: 2126-2127.

Jones, R.T. 1987. Drugs of abuse profile: Cannabis. Clinical Chemistry Vol. 33(Suppl.): 72B-81B.

Kozel, NJ. & Adams, E.H. 1986. Epidemiology of drug abuse: An overview. Science Vol. 234: 970-974.

Larsen, J & Fogerson, R. 1988. Nonsteroidal anti-inflammatory drug in TDx assays for abused drugs. Clinical Chemistry Vol. 34: 987-988.

Litovitz, T.L.; Schmitz, B.F. & Holm, K.C. 1999. Annual report of American Association of Poison Control Centers National Data Collection System. American Journal of Emergency Medicine Vol. 7: 495-545.

Long, K.L. 1989. The discovery process in drug use testing litigation. Journal of Forensic Science Vol. 34: 1454-1470.

Lund, A.K.; Preusser, D.R; Blomberg, R.D. & Williams. A.F. 1988. Drug use by tractor-trailer drivers. Journal of Forensic Science Vol. 33: 649-661.

Mason, A.P. & McBay, A.J. 1985. Cannabis: Pharmacology and interpretation of effects. Journal of Forensic Science Vol. 30: 615-631.

McAuliffe, W.E.; Rohman, M.; Santangelo, S.; Feldman, B.; Magnuson. E.; Sobol, A. & Weissman, J. 1986. Psychoactive drug use among practicing physicians and medical students. New England Journal of Medicine Vol. 315: 805-810.

McAuliffe, W.E.; Wechsler, H.; Rohman, M.; Soboroff, S.H.; Fishman, P.; Toth, D. & Friedman, R. 1984. Psychoactive drug use by young and future physicians. Journal of Health and Social Behavior Vol. 25: 34-54.

McBay, A.J. 1988. Interpretation of blood in urine cannabinoid concentrations. Journal of Forensic Science Vol. 33: 875-983.

McBay, A.J. & Hudson. P. 1987. Cost-effective drug testing. Journal of Forensic Science Vol. 32: 575-579.

Mello, N.K. & Mendelson, J.H. 1985. Operant acquisition of marijuana by women. Journal of Pharmacology and Experimental Therapeutics Vol. 235: 162-171.

Mikkelsen, S.L. & Ash. K.O. 1988. Adulterants causing false-negatives in illicit drug testing. Clinical Chemistry Vol. 34: 2333-2336.

Mulé, S.J. & Casella, G.A. 1988. Rendering the "poppy-seed defense" defenseless: Identification of 6-monoacetylmorphine in urine by gas chromatography-mass spectroscopy. Clinical Chemistry Vol. 34: 1427-1430.

Needleman, S.B. & Romberg, R.W. 1989. Comparison of drug abuse in different military populations. Journal of Forensic Science Vol. 34: 948-857.

NIDA 1998. Mandatory guidelines for federal workplace drug testing programs: Final guidelines. Notice. Federal Register Vol. 53(69): 11970-11990.

NIDA. 1986. Consensus summary of interdisciplinary approaches to the problem of drug abuse in the workplace. Rockville, Maryland: NIDA.

Noel Dunivant & Associates. 1985. Drug Testing in Major U.S. Corporations: A Survey of the Fortune 500. Raleigh, North Carolina: Noel Dunivant and Associates.

Ostrov, E. & Cavanaugh, J.L. 1987. Validation of police officer recruit candidates: Self-reported drug use. Journal of Forensic Science Vol. 32: 496-502.

Parish, D.C. 1998. Relation of the pre-employment drug testing result to employment status: A one year follow-up. Journal of General Internal Medicine Vol. 4: 44-47.

Pearson. S.D.; Ash, K.O. & Urry, F.M. 1989. Mechanism of false-negative urine cannabinoid immunoassay screens by Maine eye drops. Clinical Chemistry Vol. 35: 636-638.

Peterson, K.W.; Austin, R.L.; Gawin, F.H. & Long, K. 1987: Panel discussion: Counseling, treatment, and employer response. Clinical Chemistry Vol. 33(Suppl.): 101B-105B.

Schuster, C.R. 1987. The United States drug abuse scene: An overview. Clinical Chemistry Vol. 33(Suppl.): 7B-12B.

Substance Abuse Testing Committee 1988. Clinical issues in urinalysis of abused substances. Clinical Chemistry Vol. 34: 605-632.

Taggart. R.W. 1989. Southern Pacific: A success story. Forensic Drug Testing Newsletter December.

Unsigned. 1989. New study indicates utility of pre-employment drug tests. Clinical Chemistry News March: 8.

Walsh, D.C. & Hingson, R.W. 1985. Whom to refer employees for treatment of drinking problems. Journal of Occupational Medicine Vol. 21: 745-752.

Warner. A. 1989. Interference of common household chemicals in immunoassay methods for drugs of abuse Clinical Chemistry Vol. 35: 648-651.

Whitney, J. 1989. Drug use patterns found in criminaI justice and workplace drug testing program. California Association of Toxicology Newsletter Fall: 21-28.

Yesavage, J.A.; Leirer, B.O.; Denari, M. & Hollister, L.E. 1985. Carry-over effects of marijuana in aircraft pilot performance: A preliminary report. American Journal of Psychiatry Vol. 142: 1325-1329.

TABLE I TIME INTERVALS FOR DETECTION OF DRUGS IN URINE AFTER USE AT CUTOFF CONCENTRATIONS
NIDA-Recommended Drug Tests	Days Detectable After Use
Marijuana metabolite
Single use	<7
Chronic abuse	<30
Cocaine metabolite	<3
Opioid metabolite	<2
Phencyclidine	<7
Amphetamines	<2
Others
Alcohol	<1
Barbiturates	<2
Phenobarbital	<7
Benzodiazepines	<3
Methaqualone	<7
Methadone	<4