Title VII Discrimination for AIDS and Marijuana in Biochemical Testing

The acquired immune deficiency syndrome (AIDS) and drug abuse in the workplace are the subjects of relentless media and government scrutiny.1 As of early 1987, public health officials had reported over 30,000 cases of AIDS in the United States. 2 By the end of 1987, at least one million Americans probably had been exposed to the fatal disease. 3 The associated health insurance costs 4 and productivity losses 5 will have a significant impact on the private sector. Likewise, the Research Triangle Institute estimates that substance abuse on the job costs the economy as much as forty-six billion dollars in 1980.6 Federal experts estimate that between ten percent and twenty-three percent of workers use drugs on the job.7 In response to mounting publicity and concern, increasing numbers of private employers have implemented biochemical testing to screen employees for drug use and for the presence of AIDS.8 Employers and insurers will probably also make great use of similar tests for AIDS in the coming years. Courts and employers, however, should interpret these biochemical tests with caution. Although facially neutral, biochemical tests may discriminate against groups protected by Title VII of the Civil Rights Act of 1964.9 Title VII, in addition to banning intentional discrimination, prohibits facially neutral employment practices that have a disparate impact on persons of a particular race, national origin, sex, or religion.10 Under the disparate impact theory, a plaintiff need only prove a significant disparate impact to establish a prima facie claim; an intention to discriminate is not a necessary element. 11

Once the plaintiff meets this burden, the employer must justify the employment practice. 12 Judicial decisions dealing with disparate impact may be separated into two categories: those dealing with selection tests and those dealing with selection policies. This distinction is important: if biochemical testing is classified as a selection policy, an employer need only articulate a single nondiscriminatory reason for the policy. Selection tests, however, are given more stringent scrutiny by reviewing courts.1 3 Because biochemical tests are to some degree inherently inaccurate, it is possible that they discriminate on a chemical level.' 4 Thus, biochemical tests have the same potential to discriminate as other professionally developed tests. Consequently, biochemical tests are most appropriately analyzed under the testing strand of disparate impact doctrine. This note argues that a disparate impact analysis must distinguish between the accuracy of a biochemical test and the policy the test may be intended to enforce.15 An employer seeking to justify a discriminatory biochemical test should follow a two-part method of validation. First, the employer should offer specific evidence verifying the accuracy of the test. The best way to ensure the accuracy of biochemical testing is to follow initial screening tests with a confirmation test. 16 The employer who verifies the accuracy of a biochemical test greatly curtails the potential for biochemical discrimination.

Second, after establishing the accuracy of the test, the employer should validate the biochemical test under the Uniform Guidelines on Employee Selection Procedures (the Uniform Guidelines) as promulgated by the Equal Employment Opportunity Commission (EEOC).' 7 Part I of this note reviews the inherent inaccuracies of biochemical testing, suggesting that screening tests must be followed by confirmation tests.' 8 Part II reviews the disparate impact theory of employment discrimination. 19 Part III suggests how biochemical testing might be implicated in disparate impact analysis.20 Part IV discusses the different standards of validation under the two strands of disparate impact doctrine and distinguishes employee selection tests from employee selection policies that may be implemented through testing.2' The section argues that courts must require employers to justify biochemical tests on the basis of their accuracy and job-relatedness. Part V proposes a reproducible standard for validating biochemical tests using the Uniform Guidelines. The section concludes that employers must ensure the accuracy of biochemical tests by using high cutoff scores and confirmation tests.22 I. THE INHERENT INACCURACY OF BIOCHEMICAL TESTING The validity of biochemical testing originates in the often fallible objectivity of scientific measurement. In order to consider more fully how a court should treat a biochemical test under a disparate impact analysis, this section first reviews the inherently inaccurate nature of biochemical testing.

A. Screening Tests. Most biochemical tests begin with a screening test that is highly sensitive, but not highly specific. 23 Common screening tests for marijuana in the workplace are the radioimmunoassay (RIA) and the enzyme-multiplied immunoassay technique (EMIT). 24 A commonly used AIDS screening test is the enzyme-linked immunosorbent assay (ELISA). 2 " One problem with screening tests is that they are inferential: they only detect the chemical by-products, or metabolites, of drugs or the AIDS virus, not the drug or virus itself. 26 Other substances have the same molecular shape as the by-products of marijuana and the AIDS virus; screening tests cannot differentiate between these "cross-reactive" impostors and the by-products of the AIDS virus or drugs. Cross-reactive substances fit the molecular "keyhole" of a biochemical test and cause a "false positive."' 27 For example, one form of the EMIT will falsely register positive for marijuana when a common pain reliever, ibuprofen, is present in urine.28 The high sensitivity of marijuana screening tests is also problematic. A urine test detects minute quantities of marijuana by-products in the urine. Whether the amount detected represents a "positive" result depends solely on the arbitrary setting of a "cutoff level."'29 When metabolites detected are in excess of this predetermined cutoff level the result is considered positive; a very low cutoff level may yield a positive result for marijuana over a month after ingestion.30 Moreover, an EMIT test with a low cutoff level may. indicate a positive result when an employee has had only passive exposure to marijuana; an employee could test positive by unintentionally inhaling the marijuana smoke of those around her, 31 just as non-smokers passively inhale the cigarette smoke of others. Further, due to the fat-soluble nature of marijuana metabolites, urine screening tests may continue to show a positive result when metabolites leach out of fat cells long after actual intoxication. 32 A marijuana screening test, therefore, cannot measure impairment; it can only inferentially measure ingestion.

Thus, courts and employers cannot rely on screening tests alone as an indicator of drug use that might compromise job performance. Because of cross-reactivity and oversensitivity, positive results in a screening test will often have no relationship to an impairment or, more importantly, job performance. One way to improve the accuracy of screening tests is to raise the cutoff level, but the problem of cross-reactivity remains. Only confirmation tests come close to eliminating both oversensitivity and cross-reactivity. B. Confirmation Tests. Confirmation tests eliminate much of the inaccuracy in biochemical testing and are more specific than screening tests.33 They tend to locate false positives and confirm the initial readings of a screening test only when a true positive is present. The most effective confirmation test for drug testing is gas chromatography-mass spectrometry (GC-MS), a test so precise that it is often called "molecular fingerprinting."3 4 The most common test to confirm an initial positive for the AIDS virus is the Western blot assay. 35

Employers, however, do not always use confirmation tests because they usually are more expensive and more labor-intensive than screening tests.3 6 Nonetheless, the use of confirmation testing can only benefit an employer. A truly job-related, nondiscriminatory test will deter frivolous litigation and contribute to safety, efficiency, and profit maximization in the workplace.37 Many courts have required that screening tests be followed by confirmation tests.38 II. THE DISPARAT]E IMPACT THEORY OF EMPLOYMENT DISCRIMINATION Congress intended that Title VII eliminate employment discrimination on the basis of race, sex, religion, or national origin.39 In the seminal case of Griggs v. Duke Power Co., 40 the Supreme Court set forth a theory of employment discrimination under Title VII known as "disparate impact. 41 The disparate impact theory seeks to remedy the effect of past discrimination and to remove "artificial, arbitrary, and unnecessary barriers to employment."'42

Since the Griggs decision was handed down, however, disparate impact doctrine has been less than coherent. Issues as basic as what weight a court should give the Uniform Guidelines remain unresolved. 43 Moreover, lower courts recognize a variety of methods for establishing a prima facie claim of disparate impact.44 More important, Griggs itself uses several different phrases to describe the standard of an employer's business justification.4 5 The most recent point of contention involves the question of whether the scope of disparate impact testing includes subjective hiring practices. 46

In Griggs, the Supreme Court ruled that employment practices having a disparate impact on a protected class violate the prohibitions of Title VII unless the employer can show that the selection device is "job related." 47 Before the effective date of Title VII in 1965, Duke Power hired blacks only to work in its labor department.4 8 After the enactment of Title VII, Duke Power abandoned its policy of overt discrimination and substituted testing procedures. The company required that applicants to departments other than labor have a high school diploma or that they achieve satisfactory scores on written aptitude tests. 49 The plaintiffs argued that although these requirements applied equally to blacks and whites, they disproportionately excluded blacks from hiring and promotion. 0 The Court found that Duke Power was unable to justify the employment practices.51 Griggs and its progeny 52 set forth the core of the current disparate impact doctrine. There are two steps that must be satisfied. First, a plaintiff must make a prima facie showing that a facially neutral selection procedure has a disparate impact on his or her protected group. 53 To establish a significant discriminatory effect, plaintiffs usually produce statistical data. For example, the Griggs Court relied on evidence that 12% of black males and 34% of white males held high school diplomas in North Carolina, and on an unrelated EEOC finding that only 6% of blacks, but 58% of whites, had passed the disputed tests. 54

The statistical approach used in Griggs is instructive but not binding. Although the Supreme Court extensively discussed the many issues associated with statistical evidence of employment discrimination in Hazelwood School District v. United States, 55 it has yet to endorse a particular statistical method for showing a prima facie claim of disparate impact.5 6 The lower federal courts recognize several methods of statistical proof, most notably the two-tailed test of standard deviation and the 80% rule.57 Once a plaintiff has established a prima facie claim of disparate impact, the second step of the Griggs analysis follows. In this step, the defendant assumes the burden of justifying the employment practice in question.58 What justifies an employment practice is not entirely clear, although Griggs appears to require that an employer justify an employment practice by showing that it is "job-related." 59 In subsequent decisions, however, the Court has emphasized the Griggs language requiring an employer to show the "business necessity" of an employment practice.60 In cases not involving objective tests, the Court has not required strict validation at all. 61

The Griggs Court accorded "great deference" to the old EEOC Guidelines as a method for validating employee selection tests; 62 Duke Power was required to "validate" the test under a method based primarily on the old EEOC Guidelines. 63 More recently, the Supreme Court has retreated from strict reliance on the EEOC Guidelines." The Guidelines as they existed when the Court decided Albemarle Paper Co. v. Moody imposed a harsh burden on employers;65 the Supreme Court and lower federal courts have explored other, more workable validation methods for tests and for non-test devices, typically relying indirectly on the Guidelines. 66 It is important to note, however, that even though the method of validation fluxes, the requirement that validation occurs does not. III. BIOCHEMICAL DISCRIMINATION Under the disparate impact theory a plaintiff need not prove why a test is discriminatory.

To establish a prima facie case, a plaintiff must only make the statistical showing required in that particular jurisdiction. As this note discusses below, the disparate impact analysis applies to biochemical tests. 67 Thus, when an employer uses a biochemical test as a pass/fail barrier to employment, a plaintiff need only show that the test excludes a disproportional number of a protected class, not how the test excludes them. Once a plaintiff makes this showing, the employer must articulate a business justification for the biochemical test. Because of a biochemical test's potential inaccuracy, however, an employer must also prove the accuracy of a biochemical test. A biochemical test might have a disparate impact on the members of a protected class for a number of reasons, but the most troubling is that the test itself may subtly discriminate regardless of the employment policy it enforces. For example, it is possible that some substance peculiar to the physiology of a given race may be cross-reactive and thereby cause false positives in common screening tests like the RIA or EMIT. 68

According to one study, low-risk populations such as caucasian women and Asians might have a false positive rate as high as 88.7%.69 The study theorizes that the Western blot confirmation tests will be accurate in 71.8% of cases.70 Biochemical tests may also reflect socioeconomic patterns of drug use and AIDS infection. It is well-documented that AIDS is more prevalent among Blacks and Hispanics than among other racial groups. 71 Because AIDS strikes certain groups with greater frequency, an AIDS screening test might be discriminatory in effect if what the test purports to measure is not job-related. 72 Also, in certain locales, marijuana use rates among a minority group may exceed the rates of other groups. Non-user minorities in these locales have a greater likelihood of passive exposure to marijuana smoke.73 These examples are only possible explanations for a statistical instance of disparate impact. Under Griggs, however, the cause of the disparity would be irrelevant. Confirmation tests greatly reduce the potential for biochemical discrimination present at the screening test level.

The GC-MS test for marijuana effectively eliminates most cross-reactive substances. 74 In addition, if the GC-MS test is preceded by a screening test that raises the cutoff level for positive results to 100 nanograms per milliliter, the probability of detecting passive exposure or off-duty use is minimized. 75 Likewise, if cutoff levels for the ELISA test for AIDS are raised in low-risk populations and the ELISA test is confirmed with the Western blot assay, the numbers of false positives decrease. 76 Even if an AIDS biochemical test is positive, however, an employer would need to prove that a positive test result is job-related. Such proof would be difficult unless the patient was afflicted with debilitating AIDS encephalopathy or brain disease. 77 A positive AIDS test without such accompanying manifestations would probably not be sufficiently job-related since the majority of research indicates that the disease cannot be transmitted by casual contact.78 Only in jobs involving health care might an AIDS test be job-related, although an employer could make an argument that AIDS testing is job-related because of the potential rise in health insurance premiums. IV. BIOCHEMICAL TESTS AS EMPLOYEE SELECTION DEVICES Griggs and its progeny establish two strands of disparate impact doctrine: the selection policy strand and the professionally developed testing strand. This section discusses both strands and explains why biochemical tests comfortably fit within traditional disparate impact definitions of employee selection tests. Finally, this section distinguishes a test from the policy it implements. When the accuracy of the test itself is called into question, an employer seeking to prove its job-relatedness must offer evidence of the test's accuracy and validate the test.

A. The Selection Policy Strand. In several decisions, the Supreme Court has applied the disparate impact doctrine to non-test selection devices. These decisions typically involve an employer's policy requiring employees to meet specified criteria or possess certain qualities. These decisions differ from testing decisions because the method of assessing or measuring the employer's requirement is not in question; it is the policy that the employer must justify. Consequently, in these decisions, the Court has not required validation under the Uniform Guidelines. In Dothard v. Rawlinson, 79 the Court considered a non-test selection policy-Alabama statutes that established minimum height and weight requirements for applicants for the position of prison guard.80 The requirements would have excluded 41.13% of American women while excluding less than 1% of men.81 The Court did not require proof of the accuracy of a weight scale or of a yardstick; these were implicitly accepted as valid and accurate. The Court scrutinized the policy of requiring a minimum weight and height, not the methods of measurement. The employer argued that the policy selected applicants on the basis of strength. 82

The Court held that the employer had not shown the job-relatedness of the selection policy for two reasons. First, the Court reasoned that the employer could have used an alternative test that directly measured strength and defensive capacity.8 3 Second, in a footnote, the Court stated, "A discriminatory employment practice must be shown to be necessary to safe and efficient job performance to survive a Title VII challenge."' 84 The Court did not require validation under the Uniform Guidelines, but the alternative test rationale is considered in the Guidelines. 85 In New York City Transit Authority v. Beazer, 86 the Court also considered a non-test selection policy. In Beazer, the employer prohibited the hiring or promotion of workers using any narcotic,87 including the legal therapeutic agent methadone, prescribed to treat heroin addiction. 88 When the employer discovered that employees were enrolled in methadone maintenance programs, the employees were dismissed. 89 The employer also used urine testing as a method of discovering drug abuse on the job. 90 The urine test was one mode of implementing what was primarily a selection policy rather than a pass/fail barrier to employment. The Transit Authority justified the blanket policy as necessary for the safe and efficient performance of work that could potentially endanger the public.9' The Supreme Court agreed that the "legitimate employment goals of safety and efficiency" justified the employment policy.92 The selection policy in Beazer implicated Title VII because the plaintiffs produced statistics showing that more Blacks and Hispanics than whites are involved in methadone maintenance programs. 93 The Beazer Court, however, rejected the statistical basis of the plaintiffs' claim.

94 Arguably, the plaintiffs did not establish a prima facie claim of disparate impact and the Court did not formally address the issue of proving the job-relatedness of the policy. If Beazer did not present a claim of disparate impact, the language the Court devoted to the job-relatedness issue95 is dicta. In any event, the Court's dicta indicate a less demanding standard for proving job-relatedness in the non-test selection policy context. Beazer may settle the question of whether the policy of excluding drug users from employment is related to the job. But Beazer does not address urine testing; the proper standard for proving the job-relatedness of a potentially inaccurate test is still validated under the Uniform Guidelines. B. The Testing Strand. Griggs and Albemarle Paper v. Moody 96 are the controlling decisions When the selection device is a test. Although in Griggs and Albemarle the Court relied on the EEOC Guidelines, in subsequent decisions the Court has shown less deference to the Guidelines. 97 Nevertheless, Griggs and Albemarle are still good law. Thus, while no longer entitled to great deference, the Uniform Guidelines continue to guide courts and employers when the selection device in question is a test. Washington v. Davis 98 is the only Supreme Court decision involving a selection test that does not require a demanding standard of validation.

In Davis, the District of Columbia police department required applicants to pass a written examination measuring verbal and communicative skills. 99 The police department argued that a positive correlation between test results and performance in a training program satisfied the job-relatedness requirement. 1°0 The Court found this justification sufficient to validate the test. 10' The Davis holding, however, is of limited value in the Title VII context. The Court grounded its decision on the plaintiffs' constitutional claim and not Title VII.102 Most lower courts have limited the application of Davis, requiring more than correlation between test scores and success in training for validation.103 C. Biochemical Tests as Selection Tests. Disparate impact decisions of the testing strand have typically involved paper-and-pencil tests such as those in Griggs and Albemarle. Supreme Court decisions, however, have in no way limited employee selection devices to paper-and-pencil testing;a°4 the Court has consistently suggested a broad definition of testing. In Griggs, the Court referred to "testing or measuring procedures, "105 suggesting that the scope of selection testing extends beyond psychological aptitude or ability tests to other measuring procedures. Moreover, in Albemarle the Court expanded the meaning of testing to include employee "tests or selection devices," 10 6 again implying that employee selection tests were not limited to paper-and-pencil examinations. In Albemarle Justice Stewart also impliedly affirmed the view that the term "testing devices" includes both "testing [and] measuring procedures."107 F

Furthermore, the Supreme Court has shown some deference towards selection tests under the EEOC Guidelines. The Guidelines broadly define tests or selection procedures to include measurements of "physical" requirements.10 8 The Guidance Document to the Guidelines also supports a broad interpretation of testing, defining employee testing as "selection procedures." 10 9 Selection procedures include "physical requirements" and "physical job requirements."' 10 Biochemical tests, therefore, come within current Supreme Court definitions of selection testing. Because of their inherent inaccuracy, biochemical tests may contribute to unintentional discrimination just as easily as paper-and-pencil aptitude tests. 111 D. The Relationship of Tests and Policies. In general, a test seeks to measure a specific quality or characteristic. A measurement of a person's vision, height, or weight is a test, of sorts. The connection, however, between the quantitative measurement and the inference drawn from the measurement is so close that one cannot conceive of the two as anything but the same thing. This sort of test is self-validating. The tests in Dothard v. Rawlinson " 2-weight and height-were self-validating. The Court did not require proof of the validity of a weight scale or of a yardstick. These tests were implicitly accepted as valid and accurate. The Court scrutinized the policy of requiring a minimum weight and height, not the methods of measurement.

Other tests and the inferences drawn from them are not so closely related. Before a test such as a scholastic aptitude test achieves credibility and efficacy it must be psychologically valid. This validity is often dependent on certain underlying cultural and social assumptions about what is considered intelligence or academic skill. In the employment context, they are typical assumptions about safety, efficiency, and profit maximization. These assumptions are why the Griggs and Albemarle Courts were concerned with the validity of the paper-and-pencil tests. Not only must the quality measured be job-related, but the method of measuring that quality must be accurate. Otherwise, job-relatedness becomes irrelevant. Most importantly, a test is something quite distinct from the policy it implements. When the two are used in tandem, a court must regard the test and the policy as two discrete elements; the test must be accurate, and the policy it enforces must be job-related.

Typically, validation of the test-proof of its accuracy will also justify the policy as one related to the job. This is especially true when an employer uses the content variety of validation under the Uniform Guidelines. In the case of biochemical tests, however, the potential for inaccuracy compels a two-part analysis: first, an employer must prove the accuracy of the test, and second, the employer must prove the validity of the policy the test implements. V. VALIDATING BIOCHEMICAL TESTS. A. The Advantages of Validation Under the Uniform Guidelines. Because of the potential inaccuracies, courts should require employers to validate biochemical tests under a system of demanding and reliable rules to ensure their accuracy and job-relatedness. In Davis, the Court stated: "[T]here is no single method for appropriately validating employment tests for their relationship to job performance."'1 13 In spite of the Davis aberration, Griggs and Albemarle are binding in the testing context. The Uniform Guidelines must serve as at least the basis for validating biochemical tests. The Uniform Guidelines offer several advantages as a method of validation. For example, the Guidelines provide administrative expertise and expert study in the technical and specialized area of testing.

The Guidelines are predictable and reliable; courts can point to the same objective standard in each case of biochemical testing. Once an employer has validated a biochemical test under the Guidelines, she can use the test with confidence. Without a consistent method, the employer would have less confidence in her selection device and might forego drug or AIDS testing altogether. A truly nondiscriminatory test can increase the productivity and safety of the employer's business. Other advantages are administrability and efficiency. Courts and employers alike can consult the Guidelines rather than groping for some vague notion of safety or efficiency. If courts allow only a safety and efficiency standard and do not focus on the accuracy of the test, the potential for discrimination on the chemical level will always be present. Tests of confirmation must be used in addition to a validation study to eliminate the possibility of chemical discrimination. Finally, the Guidelines can make validation easier for an employer. The Guidelines allow "transportation" of the validity studies of others.114 That way an employer would not have to conduct his own expensive study if another employer had already done a similar study. Also, the Guidelines allow an employer to replace the challenged test with a nondiscriminatory alternative procedure. B. Validation: A Paradigm. An employer should conduct a validation study before implementing a biochemical test as a selection device. If a court finds that the employer has not conducted a validation study, it should compel validation under the Uniform Guidelines. A court that is assessing the sufficiency of an employer's validation study should begin by determining whether the employer retained the necessary data and records. First, a court must require evidence of the accuracy of the test-proof of the use of confirmation tests, records of cutoff levels, and other technical elements of the test. Second, a court should require records of test results.' 15 The results, should be tabulated for minorities and non-minorities. Third, a court should require the employer to retain data on the safety sensitivity of the employees' jobs. 116 Next the court should determine whether the alternatives of transportation or modification are available. Transportation refers to the borrowing of a validity study conducted by another employer, provided the tests and job categories are sufficiently similar.

117 Modification refers to the simple replacement of the discriminatory selection test with a less discriminatory alternative. 11e If transportation and modification are unavailable as alternatives, then a validation study will be necessary to rebut the prima facie case. The validation study should be of the criterion-related variety.' 19 When validation is necessary, the court should require initially that the employer introduce evidence of the accuracy of the test. The method of obtaining the sample should be noted, as well as the cutoff score used to call a result "positive." 120 A cutoff score set too low may invalidate the job-relatedness of the test by introducing the possibility of detecting cross-reactive compounds or the off-duty use of marijuana in the case of urine testing. 121 The court should ascertain whether the employer used a more specific test to confirm the screening test. 122

The employer should undertake a job analysis for each affected job category. This analysis should describe duties, safety-sensitivity, work behaviors, and work outcomes of the job. 123 Furthermore, specific criteria for each job category should be named and rated, including but not limited to production rate, error rate, tardiness, absenteeism, and length of service.124 These normative categories should be tabulated for both minorities and non-minorities so that the requirement for differential validation can be met. 2 5 Steps should be taken to formalize these criteria on standardized forms in an effort to minimize supervisor (i.e., "rater") bias.

A court might then determine whether employees submitted to a voluntary program of random drug testing. The employees chosen should reflect a random cross-section or representative sample of ethnic and racial categories as well as positions in the line of career progression. 127 Statisticians should assess the correlation between job performance and marijuana levels in the urine or the numerical value of the AIDS test. If a positive correlation between measurements exists, then the employer may present these findings to the court as evidence that the biochemical test is in compliance with the Guidelines.

Even if a validation study establishes the accuracy of a biochemical test, the test still may not be job-related. This is especially true in the case of AIDS testing; the majority of research indicates that, contrary to popular wisdom, one cannot contract AIDS through casual contact. 129 An employee infected with AIDS will perform at a similar physical level until the later stages when brain disease develops. Therefore, even with high cutoff levels and the Western blot assay, AIDS testing may not be job-related except in certain health-related professions and occupations such as the armed forces, where blood donations are required. Validation eliminates the possibility of discrimination and allows the necessary testing in health-sensitive positions. If an employer proves the accuracy of a urine test for drugs, the employer can probably justify the test on the grounds of safety and efficiency. This rationale is persuasive in safety-sensitive positions, but in non-safety-sensitive positions, a urine test may not be directly related to job performance. The oversensitivity of urine tests makes them especially problematic in non-safety-sensitive positions. An employee could test positive weeks after ingestion of the drug. 130 Arguably, what an employee does during his off-duty hours is a private matter and is not related to job performance. On the other hand, an employee's off-duty conduct may reflect on his business, perhaps damaging goodwill. The inquiry into the job-relatedness of urine testing is thus fact-specific and can proceed only when the employer has verified the accuracy of the urine test.

Employee urine testing should increase disparate impact litigation under Title VII. Socioeconomic patterns of drug use coupled with passive inhalation and the potential for cross-reactivity may cause some ethnic groups to suffer unduly from biochemical drug testing. Given the proper use of statistical data, plaintiffs may state a prima facie case and force employers to validate biochemical tests. Cutoff levels set too low will detect passive exposure, thereby differentially excluding some from hiring or promotions. The proliferation of AIDS testing in the employment setting also should bring an increase in Title VII litigation. In the first instance, Blacks and Hispanics are afflicted with AIDS at higher rates than whites. Simply because individuals test positive for the AIDS virus does not imply that performance on the job will be affected or that unsafe conditions will result. Only in the late stages of AIDS would emaciation or encephalopathy impair job performance.

Likewise, employees with AIDS would not create unsafe conditions for other employees or the public since the virus is not contagious through casual contact. Furthermore, most test results of the ELISA screen for AIDS will be falsely positive in low-risk populations. If an employer tested protected members of a low-risk population and these applicants were excluded on the basis of unconfirmed ELISA screens, then a claim for disparate impact could follow. The heightened potential for Title VII litigation underscores the importance of defenses for employers. If biochemical testing is classified as an employee selection policy, then the employer needs simply articulate a single nondiscriminatory reason why the policy promotes safety and efficiency. If biochemical testing is classified as an employee selection test, then the employer must either validate, transport validation, or modify the protocols for biochemical testing. The case law, the EEOC Guidelines, and the Guidance Document all support the conclusion that biochemical testing should be classified as an employee selection test. Furthermore, analysis of biochemical testing as selection testing offers several policy advantages. First, validation studies will minimize the effects of discriminatory biochemical tests and discriminatory uses of valid biochemical tests. Second, with ready knowledge of a biochemical testing protocol's validity, an employer could make employment decisions more confidently. Finally, validation serves the goals of safety, efficiency, and profit maximization. A valid selection test can benefit all; the employer is allowed to pursue valid business goals while society protects the rights of all categories of citizens.