Sylvia HAYES, Plaintiff-Appellee, v. SHELBY MEMORIAL HOSPITAL, Defendant-Appellant

726 F.2d 1543, 34 Fair Empl. Prac. Cas. (BNA) 444, 1984 U.S. App. LEXIS 24463, 33 Empl. Prac. Dec. (CCH) 34,219

• Court: Court of Appeals for the Eleventh Circuit
• Decided: March 16, 1984
• Docket: 82-7296
• Status: Published

Opinion

TUTTLE, Senior Circuit Judge:

Historically, an effective means for employers, legislatures, and courts to limit the equal employment opportunities of women was to restrict their employment out of a professed concern for the health of women and their offspring. See, e.g., Muller v. Oregon, 208 U.S. 412, 28 S.Ct. 324, 52 L.Ed. 551 (1908). 1 In this appeal we are asked to determine whether a hospital can fire an x-ray technician when she becomes pregnant to protect the pregnant employee’s fetus from potentially harmful radiation, and to protect the hospital’s finances from potential litigation. We affirm the district court’s finding that under the circumstances of this case Shelby Memorial Hospital violated the Pregnancy Discrimination Act of 1978 when it fired Sylvia Hayes from her position as an x-ray technician immediately upon learning of Hayes’s pregnancy.

I. BACKGROUND

On August' 11, 1980, Shelby Memorial Hospital (the “Hospital”) hired Sylvia Hayes, a certified x-ray technician, to work the 3-11 p.m. shift with one other technician in the Hospital’s radiology department. Two months later, Hayes was fired after she informed her supervisor that she was pregnant. The supervisor fired Hayes after consulting Dr. Cecil Eiland, the Hospital’s then recently appointed radiology department medical director and radiation safety director, who recommended that Hayes be removed from all areas in which ionizing radiation was being used. The Hospital claims that it fired Hayes because it was unable to find alternative employment for her.

Following her dismissal from the Hospital, Hayes filed suit in the United States District Court for the Northern District of Alabama, charging that the Hospithl had denied Hayes her equal employment opportunities as guaranteed by Title VII of the 1964 Civil Rights Act, 42 U.S.C. § 2000e et seq., as well as her constitutional and civil rights guaranteed under 42 U.S.C. § 1983. The Hospital defended its dismissal of Hayes on the grounds, among others, of “business necessity” and “bona fide occupational qualification.” Following a nonjury trial the district court concluded that the Hospital’s “abrupt termination” of Hayes violated both Title VII and § 1983. The district court awarded Hayes modest damages totalling less than $8,000.00, and entered final judgment on the award, from which the Hospital took this appeal.

II. DISCUSSION

This is a case of first impression in the Eleventh Circuit, and one of only a few .cases of this nature to reach a court of appeals. 2 The issues raised by this case are (1) What is the proper legal framework for analyzing a case in which an employer fires a pregnant woman because of the employer’s belief that the woman’s continued employment presents a substantial hazard to her fetus? (2) Does an employer’s desire to avoid or minimize potential liability for fetal exposure to radiation constitute an affirmative defense to a Pregnancy Discrimination Act claim? (3) Did the Hospital meet its burden of proving an affirmative defense to its discriminatory policy?

A. The Legal Framework

We start our analysis with Title VII and the- Pregnancy Discrimination Act, 42 U.S.C. § 2000e(k). Prior to passage of the Pregnancy Discrimination Act, the Supreme Court had held in General Electric Co. v. Gilbert, 429 U.S. 125, 97 S.Ct. 401, 50 L.Ed.2d 343 (1976), that Congress did not include discrimination on the basis of pregnancy within Title VII’s definition of gender based discrimination. By adopting the Pregnancy Discrimination Act in 1978 as an amendment to Title VII, Congress in effect overruled Gilbert. Under the Pregnancy Discrimination Act, discrimination on the basis of pregnancy is discrimination on the basis of sex. Thus, our analysis of this case must follow the same type of analysis used in any other Title VII sex discrimination suit.

There are three theories under which a suit of this nature may be analyzed. 3 The first applies to those situations in which an employer has engaged in “facial” discrimination. Facial discrimination occurs when an employer adopts a policy that explicitly treats some employees differently from others on the basis of race, religion, national origin, or gender (pregnancy). The only affirmative defense to facial discrimination is the existence of a bona fide occupational qualification (BFOQ), by which an employer may establish that religion, sex, or national origin (but not race) is, in the words of Title VII, a “bona fide occupational qualification reasonably necessary to the normal operation of that particular business or enterprise.” 42 U.S.C. § 2000e-2(e). The second theory applies when an employer adopts what appears to be a facially neutral policy, but one which a plaintiff contends is a “pretext” for forbidden discrimination. Both facial discrimination and pretext cases are referred to as “disparate treatment” cases.

The third theory applies when the employee concedes that the employer’s policy is neutral, but seeks to demonstrate that the policy has a disproportionate impact on a group protected from discrimination under Title VII. ■ Those cases are labelled “disparate impact” cases. An affirmative defense to a prima facie case of disparate impact is “business necessity,” which is a broader defense than BFOQ.

We analyze this case under all three theories.

B. Analysis

1. The Pretext Theory

Initially, the Hospital argues that we should analyze this case under the pretext theory as set forth in Texas Department of Community Affairs v. Burdine, 450 U.S. 248, 101 S.Ct. 1089, 67 L.Ed.2d 207 (1981). Under Burdine, an employer may “articulate” a “legitimate nondiscriminatory reason” to rebut an employee’s prima facie ease of discrimination as made out under the requirements of McDonnell Douglas Corp. v. Green, 411 U.S. 792, 93 S.Ct. 1817, 36 L.Ed.2d 668 (1973). The employee must then show that the employer’s nondiscriminatory reason is a “pretext” for discrimination. The Hospital argues that its stated reason for firing Hayes — to protect her fetus from the harmful effects of radiation— is a sufficient nondiscriminatory reason to shift the burden back to Hayes to prove pretext. The Hospital then argues that Hayes failed to prove that its reasons for firing her were a pretext for discrimination.

Prior to passage of the Pregnancy Discrimination Act, the Hospital’s Burdine argument might have had some merit, 4 because pregnancy was considered a legitimate, nondiscriminatory basis for differential treatment. See Nashville Gas Co. v. Satty, 434 U.S. 136, 98 S.Ct. 347, 54 L.Ed.2d 356 (1977); General Electric Co. v. Gilbert, 429 U.S. 125, 97 S.Ct. 401, 50 L.Ed.2d 343 (1976). That is no longer the case, however, because the Pregnancy Discrimination Act mandates that a pregnancy-based rule can never be “neutral.” 5 In other words, firing Hayes because she was pregnant is just as facially discriminatory under the Pregnancy Discrimination Act as it would be to fire her solely because she was black under Title VII. Therefore, Burdine is inappropriate here, since the Hospital admits that Hayes was fired because of her pregnancy, rather than because of some other, potentially nondiscriminatory reason. 6

2. Facial Discrimination

Plaintiff, supported by amici curiae American Civil Liberties Union Foundation Women’s Rights Project, et al., 7 contends that this is a case of facial discrimination, to which the only affirmative defense is BFOQ. Although we tend to agree that this is a facial discrimination case, to ensure complete fairness to the Hospital, we will also analyze this case under the disparate impact/business necessity theory.

First, we must establish an analytic framework. We begin by establishing a presumption that if the employer’s policy by its terms applies only to women or pregnant women, then the policy is facially discriminatory. That presumption may be rebutted, however, if the employer can show that although its policy applies only to women, the policy is neutral in the sense that it effectively and equally protects the offspring of all employees. In other words, the employer must show (1) that there is a substantial risk of harm to the fetus or potential offspring of women employees from the women’s exposure, either during pregnancy or while fertile, to toxic hazards in the workplace, and (2) that the hazard applies to fertile or pregnant women, but not to men. 8

The burden of proving a substantial risk of harm to the fetus is a threshold requirement. “This burden may not be carried by proof alone that the employer subjectively and in good faith believed its program to be necessary and effective for the purpose.” Wright v. Olin Corp., 697 F.2d at 1196. Rather, the employer must produce objective evidence of an essentially scientific nature supported by the opinion evidence of qualified experts in the relevant scientific fields. Id. The employer need not prove the existence of a general consensus on the points within the qualified scientific community. Id. at 1191. Rather, the employer carries its burden by showing that the body of opinion believing that significant risk exists is so considerable “that an informed employer could not responsibly fail to act on the assumption that this opinion might be the accurate one.” Id.

If the employer carries the threshold burden of proving significant risk of harm, it must then prove that the hazard does not also apply to the offspring of male employees. Again, scientific evidence is necessary. We note that at present there is, .even within the scientific community, a certain amount of subtle bias that has focused research on the hazardous effects of workplace substances as they pertain to reproductive health on women more so than on men. See Williams, Firing The Woman, 69 Geo. L.J. 641, 661. In those instances in which scientific evidence points to a hazard to women, but no scientific evidence exists regarding men, an employer may be allowed to adopt a suitable policy aimed only at women. As additional research on men becomes available, however, the employer must adjust its policy or risk running afoul of Title VII.

If an employer fails to overcome the burden of proving that its policy is not facially discriminatory, then its only defense is BFOQ. Under traditional analysis, the BFOQ defense is available only when the employer can show that the excluded class is unable to perform the duties that constitute the essence of the job, duties that Title VII defines as “necessary to the normal operation of the particular business or enterprise.” 42 U.S.C. § 2000e-2(e); see Dothard v. Rawlinson, 433 U.S. 321, 97 S.Ct. 2720, 53 L.Ed.2d 786 (1977); Weeks v. Southern Bell Telephone & Telegraph Co., 408 F.2d 228 (5th Cir.1969); Burwell v. Eastern Air Lines, Inc., 458 F.Supp. 474 (E.D.Va.1978), aff’d in part and rev’d in part, 633 F.2d 361 (4th Cir.1980) (en banc) (per curiam), cert, denied, 450 U.S. 965, 101 S.Ct. 1480, 67 L.Ed.2d 613 (1981). Thus, as the district court noted in this case, “[potential for fetal harm, unless it adversely affects a mother’s job performance, is irrelevant to the BFOQ issue.” Hayes v. Shelby Memorial Hospital, 546 F.Supp. 259, 264 (N.D.Ala.1982). We hold that when a policy designed to protect employee offspring from workplace hazards proves facially discriminatory, there is, in effect, no defense, unless the employer shows a direct relationship between the policy and the actual ability of a pregnant or fertile female to perform her job. 9 The Hospital never asserted that Hayes’ pregnancy, at least until the final few weeks, would interfere with her ability to perform the duties of an x-ray technician. Nor did the Hospital try to prove that a woman’s fear for potential harm to her fetus might somehow affect job performance adversely. Therefore, under traditional BFOQ analysis, the defense is inapplicable to this case.

At least one commentator, as well as amici curiae, have argued that we should resist the “predictable” temptation to allow concern about the health interests of the fetus to prompt an expansion of the BFOQ defense. Williams, Firing The Woman, 69 Geo.L.J. at 681. Williams argues that “the effect of adhering to the traditional analysis is likely to prove more beneficial to the health interests of the fetus than such an expansion” because an employer ostensibly seeking to protect fetal health will be forced to develop a gender neutral policy to avoid Title VII liability. Id. at 681-82. By adopting a general, rather than gender-specific approach, the focus is shifted to the health of offspring as affected by the exposure to toxic substances of both men and women. Id. at 682. Such an interpretation of Title VII “thereby encourages employers to address and solve what experience has shown to be a problem for workers of both sexes.” Id. We agree with the Williams approach. We believe that the Pregnancy Discrimination Act mandates that an employer treat its workers equally when it seeks to protect their offspring.

We now apply our framework to the facts of this case. 10 The Hospital in tended that its policy apply only to pregnant x-ray technicians. 11 Under our framework, that raises the presumption of facial discrimination.

As we have stated, however, the Hospital may rebut the presumption of facial discrimination. Therefore, we must examine whether the Hospital met the threshold requirement of its rebuttal case by proving that radiation from x-rays posed a significant risk of harm to Hayes’s fetus. At trial, the Hospital’s radiation safety director, Dr. Cecil Eiland, took the extreme position that any dose of radiation is “excessive” and therefore potentially harmful to a developing fetus. Dr. Eiland even stated that it would be dangerous for a pregnant woman to sunbathe in a bikini, which would expose her fetus to radiation from the sun. Although Dr. Eiland’s statement regarding the amount of radiation he considered “excessive” appears to have been hyperbolic, his statement illustrates the fact that scientists know little about the detrimental effects of even the lowest levels of radiation. The evidence at trial, however, suggests that the general scientific approach to protecting people from radiation has shown that although any amount of radiation can have a detrimental effect on humans, it is extremely unlikely in most cases that radiation below certain doses will have a detrimental effect on any particular individual. Therefore, scientists have established various maximum “safe” levels of radiation exposure. Even the “maximum safe” levels of radiation exposure are set with a large margin of error on the side of safety, so that one exposed to a “maximum” dose normally is in no real danger, at least with our present state of knowledge. Despite the margin of safety, experts caution that even within the “safe” levels, one should always limit his or her exposure to what the industry calls “ALARA” — As Low As Reasonably Attainable — to increase that margin.

At any rate, the employer’s burden of persuasion requires that it show more than simply some risk of harm to an employee’s fetus from a workplace hazard. In this case the Hospital was required to show that the level of radiation to which Hayes would normally be exposed during the period of pregnancy, assuming she was carrying out her usual duties, would pose an unreasonable risk of harm to the fetus. The district court found that the Hospital failed to demonstrate that Hayes’s fetus would be exposed to an unreasonable amount of radiation, stating, “the defendant’s termination of the plaintiff’s employment, albeit motivated by concern for the welfare of the fetus, was an unnecessarily extreme measure.” Hayes, 546 F.Supp. at 265.

The evidence in the record supports the district court’s conclusion. Much of the expert testimony at trial centered on the question of what amount of radiation exposure is “reasonable” for a pregnant woman. Those who testified more or' less agreed that standards set by the National Council on Radiation Protection and Measurements (“NCRP”) are authoritative, and are based on conservative principles that provide for a wide margin of safety. In NCRP Report No. 53, “Review of NCRP Radiation Dose Limit For Embryo And Fetus In Occupationally-Exposed Women,” which was introduced into evidence at trial, the NCRP proposes .5 rem as the maximum radiation dose to which a fetus should be exposed. (A rem is a unit of radiation, and .5 rem is also equal to 500 millirems). By contrast, the generally accepted maximum level of exposure for a nonpregnant employee is 5 rems per year.

Most employers whose employees are occupationally exposed to radiation require their employees to wear radiation badges, small dosimeters that monitor the amount of radiation to which an employee is exposed. The radiation badges are checked periodically to ensure that an employee has not been over-exposed. At Shelby Memorial Hospital all x-ray technicians wore two radiation badges — one on the chest, and one on the finger — which were checked at least once a month. During the two months she was at the Hospital, Hayes’s chest badge showed an exposure of 20 millirems per month, and her ring badge showed an exposure of 60 millirems per month. Taking the higher reading of the ring badge, which is a less accurate measure of radiation to the abdominal area than is the chest badge, and multiplying it by the eight months thát Hayes would work before taking maternity leave, reveals a total exposure of 480 milli-rems, which is within the 500 millirems limit set by NCRP. Multiplying the lower reading from the chest badge shows a total eight month exposure of 160 millirem, well within the NCRP limits.

Nevertheless, the Hospital argues that one cannot extrapolate accurately from only two months of badge readings because the total amount of exposure each month varies widely from technician to technician. The Hospital cites evidence that some of the day shift technicians occasionally received doses of several hundred millirems in one month. But the Hospital has argued consistently, in a different context, that the duties of the day and night shift technicians are quite different. Therefore, evidence that day technicians occasionally received high doses of radiation is irrelevant to Hayes’s case. The Hospital failed to show that Hayes’s duties as a night technician were likely to ever expose her to such high levels of radiation. Of course, if.Hayes’s badge readings, as continuously monitored, ever showed that she was approaching the 500 millirem mark after only five or six months, the Hospital would have been free to reevaluate her duties. We stress, however, that the Hospital would be under a heavy burden to examine all alternative possibilities for keeping Hayes employed in some capacity. 12 Thus, we hold that the Hospital failed to meet the threshold requirement that it prove that its discriminatory policy was necessary in Hayes’s case. 13

Because the Hospital failed to prove its policy was necessary we need not reach the factual issue of whether x-ray radiation affects the offspring of employees only through pregnant women, or whether similar effects can occur from exposure to males. Documents cited by amici do suggest that radiation induced mutations can pass to offspring from the male sperm, but neither party developed the evidence on this issue at trial.

Because the district court lacked the benefit of the approach we have outlined here, and therefore did not try this case using the framework we have outlined, the Hospital might argue that we should remand this case to give it a second chance to prove its firing of Sylvia Hayes was justified under Title VII. Although we believe the district court’s findings of fact were sufficient to support our holding of facial discrimination; we must be completely fair to the Hospital. Therefore, we will assume that the Hospital could rebut Hayes’s casé of facial discrimination, and show that the Hospital would, nevertheless be liable in this case, under the disparate impact theory.

C. Disparate Impact

If an employer proves that a policy applying only to women or pregnant women employees is justified on a scientific basis and is not necessary to protect the offspring of male employees, the employer has shown that its policy is neutral, in the sense that it equally protects the offspring of all employees. Nevertheless, such a “fa-daily neutral” policy clearly has a disproportionate impact on women since only they are affected by it. Therefore, even if the employer rebuts the prima facie case of facial'discrimination, the employee has an automatic prima facie case of disparate impact. Under the disparate impact theory, however, the employer is entitled to assert the defense of business necessity. See Griggs v. Duke Power Co., 401 U.S. 424, 91 S.Ct. 849, 28 L.Ed.2d 158 (1971).

One problem with business necessity as defined under traditional Title VII analysis, but applied in the context of an employer’s fetal protection program, is the requirement that the employer’s policy be related to job performance, id. at 431, 91 S.Ct. at 853, because fetal protection does not in a strict sense have anything to do with job performance. 14 For example, Hayes’s can perform her job as an x-ray technician as well while pregnant as she could beforehand. Nevertheless, the federal courts and several commentators who have considered the issue of employer fetal protection plans have recognized the need for a narrowly circumscribed business necessity defense. See Wright v. Olin Corp., 697 F.2d 1172,1188-91 (9th Cir.1982); Zuniga v. Kleberg County Hosp., 692 F.2d 986, 992 (5th Cir.1982); Williams, Firing The Woman, 69 Geo.L.J. at 687; cf. Hayes v. Shelby Mem. Hosp., 546 F.Supp. 259, 263-64 (N.D.Ala.1982). But see Burwell v. Eastern Air Lines, Inc., 633 F.2d 361, 371 (4th Cir. 1980) (en banc) (concurring opinion), cert, denied, 450 U.S. 965, 101 S.Ct. 1480, 67 L.Ed.2d 613 (1981). We agree that the defense should be allowed. 15

The employer’s business necessity defense applies automatically, just as the employee’s prima facie case of disparate impact applies automatically. That is because to reach the disparate impact stage of analysis in a fetal protection case, the employer has already proved — to overcome the presumption of facial discrimination — that its policy is justified on a scientific basis and addresses a harm that does not affect men. To add any more requirements would be to render it nearly impossible to have a fetal protection program under any circumstances. 16

Nevertheless, the employer’s business necessity defense may be rebutted by proof that there are acceptable alternative policies that would better accomplish the purposes of promoting fetal health, or that would accomplish the purpose with a less adverse impact on one sex. See Wright v. Olin Corp., 697 F.2d at 1191. The burden of rebuttal is on the employee. In other words, to avoid Title VII liability for a fetal protection policy, an employer must adopt the most effective policy available, with the least discriminatory impact possible. To require any less would be to return to the days of Muller v. Oregon.

In the case at hand, the district court found that the Hospital failed to consider acceptable alternatives that would accomplish the Hospital’s purpose with a less discriminatory impact. In particular, the district court found that Hospital officials failed to explore seriously other duties within the Hospital that Hayes could perform. Additionally, the court found that the Hospital failed to consider rearranging Hayes’s duties within the radiology department to minimize her radiation exposure. Those findings are supported by ample evidence, and thus are not clearly erroneous. Therefore, even if we were persuaded that the Hospital’s policy was justified by business necessity, we would hold that Hayes successfully rebutted the Hospital’s business necessity defense.

CONCLUSION

In sum, we have attempted to set forth a framework using traditional Title VII principles for analyzing cases in which an employer seeks to fire (or not hire) an employee to protect that employee’s offspring from some workplace hazard. Although our discussion has been lengthy, theoretical, and perhaps confusing, the effect of our analysis is fairly simple: If an employer has a fetal protection policy that applies to members of one sex only, the policy violates Title VII unless the employer shows (1) that a substantial risk of harm exists and (2) that the risk is borne only by members of one sex; and (3) the employee fails to show that there are acceptable alternative policies that would have a lesser impact on the affected sex.

In the case at hand, we hold that the Hospital failed to rebut the presumption that its abrupt firing of Sylvia Hayes when she revealed her pregnancy was the result of a facially discriminatory policy. The Hospital also failed to consider less discriminatory alternatives to firing Hayes. Therefore, we affirm the district court’s finding of liability.

The Hospital has also raised several issues regarding the district court’s computation of a modest award of damages to Hayes. Finding no clearly erroneous factual findings and that the district court applied the correct law, we affirm the damage award.

AFFIRMED.

1 . In Muller the Supreme Court upheld an Oregon statute that limited the number of hours a woman could work each day. In doing so, the court distinguished Lochner v. New York, 198 U.S. 45, 25 S.Ct. 539, 49 L.Ed. 937 (1905), which struck down a similar restriction on men’s working hours as a violation of the Contract Clause, by holding that the “difference between the sexes” justified a different rule. The Court explained the justification for upholding the Oregon statute as follows:

That woman’s physical structure and the performance of maternal functions place her at a disadvantage in the struggle for subsistence is obvious. This is especially true when the burdens of motherhood are upon her. Even when they are not, by abundant testimony of the medical fraternity continuance for a long time on her feet at work, repeating this from day to day, tends to injurious effects upon the body, and as healthy mothers are essential to vigorous offspring, the physical well-being of woman becomes an object of public interest and care in order to preserve the strength and vigor of the race.

Still again, history discloses the fact that woman has always been dependent upon man. He established his control at the outset by superior physical strength, and this control in various forms, with diminishing inten sity, has continued to the present. As minors, though not to the same extent, she has been looked upon in the courts as needing special care that her rights may be preserved ....

208 U.S. at 421, 28 S.Ct. at 326.

2 . In Zuniga v. Kleberg County Hospital, 692 F.2d 986, 992-94 (5th Cir.1982), the court addressed a pre-Pregnancy Discrimination Act firing of a pregnant x-ray technician and, using a Title VII sex-discrimination analysis, held that the Hospital failed to utilize a less discriminatory alternative to firing the technician. The court noted that the Hospital could have requested the technician to take a leave of absence pursuant to established policy. Id In Wright v. Olin Corp., 697 F.2d 1172 (4th Cir. 1982), the defendant sought to justify its exclusion of female employees of child-bearing age from several job classifications that required contact with toxic substances believed capable of causing harm to a developing fetus. The Olin court addressed at length the legal framework that should be applied to an employer’s business necessity defense predicated on protection of a fetus. Olin did not make clear, however, whether it was applying pre- or post-Pregnancy Discrimination Act principles. Although we believe Olin reaches a correct result, we have endeavored to present a clearer picture of the overall framework under which such a case should be analyzed.

3 . For an excellent discussion of the history and analytical framework of Title VII and its relation to pregnancy-based discrimination, from which much of the following discussion is drawn, see generally Williams, Firing The Woman To Protect The Fetus: The Reconciliation Of Fetal Protection With Employment Opportunity Goals Under Title VII, 69 Geo.L.J. 641, 673-682 (1981) [hereinafter, Williams, Firing The Woman].

4 . But see Wright v. Olin Corp., 697 F.2d 1172, 1185 (4th Cir.1982) (Burdine and McDonnell Douglas set forth “wholly inappropriate” scheme for analyzing pregnancy discrimination or fetal protection case).

5 . Nevertheless, the- Pregnancy Discrimination Act does not completely preclude pretext analysis in the area of employer policies designed to protect employee reproductive health. See Williams, Firing The Woman, 69 Geo.L.J. at 682-87. A plaintiff could still show that a truly neutral rule, such as one aimed generally at employee health or employee reproductive health and potentially affecting both sexes, is a pretext. Id. at 682. For ways in which a plaintiff might prove pretext in such a case, see id. at 683-85.

6 . This might be a Burdine type case if, for example, the Hospital had dismissed Hayes claiming she was “inefficient,” but Hayes had asserted that the real reason for her dismissal was her pregnancy. The court would then be required to determine if the Hospital’s stated reason for firing Hayes was a pretext for getting rid of a pregnant employee.

7 . Other amici include The Alabama Civil Liberties Union; The American Federation of State, County and Municipal Employees; The Coal Employment Project; The Committee for Abortion Rights & Against Sterilization Abuse; The Employment Law Center; Equal Rights Advocates; The International United Automobile, Aerospace, & Agricultural Implement Workers of America; NOW Legal Defense & Education Fund; The Oil, Chemical & Atomic Workers International Union; Planned Parenthood Federation of America; Reproductive Rights National Network; Women Employed; and Women’s Legal Defense Fund.

The Equal Employment Opportunity Commission has filed a separate amicus curiae brief in this case.

8 . We borrow these requirements from the showing of “business necessity” reqüired in Wright v. Olin Corp., 697 F.2d at 1190-91. Because we approach this case differently than did the Olin court, we do not label these requirements as “business necessity.” Nevertheless, our approach and that of Olin require an identical showing by an employer for the employer to prevail.

9 . Such a case should be rare. Nevertheless, it is conceivable that an employer could show that fear of fetal defects caused women as a class to become unable to perform certain jobs effectively, whereas the same problem did not exist for men as a class when they knew their jobs posed a potential hazard to their offspring.

10 . .One problem that arises at the outset of our analysis is our impression, based on the evidence in the record, that the Hospital had no “policy” in effect regarding pregnant x-ray' technicians at the time Hayes was fired. Rather, as the district court tacitly recognized, the Hospital seems to have approached the problem on an ad hoc basis. For example, two other x-ray technicians at the Hospital, in admittedly different circumstances, were allowed to retain their positions after becoming pregnant. The Hospital tried to use post hoc rationalization to differentiate the treatment of those two technicians from that of Hayes, but the Hospital conveyed the impression that it had not given the issue much thought on a continuing “policy” basis.

We mention our impression that the Hospital lacked any real policy, because it is clear that an institution with a carefully studied, narrowly tailored, written policy concerning the steps it will take if an occupationally exposed employee becomes pregnant, will be in a much stronger position than was Shelby Hospital. Such a policy can inform employees of their rights before they become pregnant, provide warning about the potential dangers of the employment, and demonstrate that the employer has attempted to strike a balance between employee safety and equal opportunity employment. In this case both the Hospital and Hayes pointed to a carefully crafted policy for pregnant x-ray technicians at the University of Alabama, Birmingham, Hospital (“UAB”). The Hospital stated that it was in a much different position than UAB, the Hospital being a much smaller institution than UAB. But the Hospital was not prevented from preparing its own policy, taking into account any peculiar circumstances that seemed appropriate. Of course, we do not intimate that any such policy (including that of UAB) would necessarily be upheld by any court. Indeed, such a policy is due for the same scrutiny we have applied here and must be supported by more than “a subjective and good faith belief” that it is necessary.

11 . We have some concern that the absence of a policy that clearly applies only to pregnant women, coupled with evidence that the Hospital employed no male x-ray technicians, could be interpreted as a policy prohibiting males and pregnant females from being employed as x-ray technicians at the Hospital. Such a policy might be neutral, because some nonrecord evi- • dence brought to our attention by, amici suggests that males of reproductive age and pregnant females faced greater radiation hazards in terms of effect on offspring than did nonpregnant females. Therefore, a policy barring all males and pregnant females from working as x-ray technicians might not be illogical, or discriminatory. Nevertheless, we doubt that the Hospital had such a policy in mind, and it certainly has never made such a contention, nor was any evidence introduced at trial on this point.

12 . See discussion under disparate impact, infra.

13 . The Hospital’s policy would probably have been ineffective in protecting Hayes’s fetus, even if the Hospital had proved the existence of a substantial danger. The evidence suggests that the greatest danger of fetal damage from radiation occurs during the earliest days of pregnancy, when a few cells in the developing embryo rapidly divide into more cells. If a single cell is affected at that time it may pass the effect to many more cells than a single cell affected at a later time. The Hospital’s policy, however, applied only after a pregnancy was discovered, which in this case was about eight weeks after conception. During the time before Hayes’s pregnancy was discovered serious injury could have occurred to her fetus, if we accept the Hospital’s contention that Hayes was exposed or could have been exposed to excessive radiation. Evidence such as this that a danger exists, but that the policy designed to redress the danger is largely ineffective in protecting the fetus, could be used to prove that such a policy was a pretext for discrimination, particularly when the policy has a harsh effect on pregnant employees (e.g., firing them).

14 . For a good example of a court’s struggle to fit the business necessity defense for a fetal protection program into a traditional definition of business necessity, see Wright v. Olin Corp., 697 F.2d 1172, 1189 (4th Cir.1981). In Olin, the court defined business necessity as “whether there exists an overriding legitimate business purpose such that the practice is necessary to the safe and efficient operation of the business.” 697 F.2d at 1188, quoting Robinson v. Lorillard Corp., 444 F.2d 791, 798 (4th Cir. 1971); see also Dothard v. Rawlinson, 433 U.S. 321, 331 n. 14, 97 S.Ct. 2720, 2728 n. 14, 53 L.Ed.2d 786 (1977). Under that definition an employer is entitled to take employee safety and the safety of legitimate business visitors into account in determining what is a business necessity. But the definition does not seem to include the safety of unborn offspring. In an attempt to reconcile the definition with the court’s conclusion that protection of offspring is a legitimate business necessity the Fourth Circuit compared the rights of an employee’s fetus vis-a-vis the employer to the rights of a business invitee or licensee vis-a-vis the employer. Olin, 697 F.2d at 1189. We believe it unnecessary to place the fetus into a legal classification developed for an entirely different aspect of the law, and therefore decline to endorse Olin’s approach of equating a fetus with a business invitee or licensee. Instead, we simply recognize fetal protection as a legitimate area of employer concern to which the business necessity defense extends.

15 . The Hospital contends that the basis for the business necessity defense in a fetal protection case such as this is the potential for litigation costs, in the form of a lawsuit by a hospital employee or her child in the event of injury caused by fetal exposure to radiation. See Zu-niga v. Kleberg County Hospital, 692 F.2d 986, 992 n. 10 (5th Cir.1982) (dicta) (“The economic consequences of a tort suit brought against the Hospital by a congenitally malformed child could be financially devastating, severely disrupting the ‘safe and efficient operation of the business.’ ”) But cf. Wright v. Olin Corp., 697 F.2d 1172, 1190 (4th Cir.1982). We cannot accept the Hospital’s contention that potential litigation costs can form the basis for the business necessity it asserts. As the district court noted, the Hospital’s position would amount to “an unwarranted extension [that] would shift the focus of the business necessity defense from a concern for the safety of Hospital patients to a focus of concern for Hospital finances.” 546 F.Supp. at 264. Here, the Hospital’s emphasis on the avoidance of litigation costs, rather than the health of its employees’ offspring, underscores our belief that the Hospital disregarded Hayes’s civil rights out of concern for the Hospital’s finances, not out of any genuine concern for Hayes’s fetus.

In today’s litigious society, the potential for litigation rests in almost every human activity. For example, every employer faces the risk that a pregnant employee will encounter a workplace activity that would not normally be hazardous to a nonpregnant employee, but which could prove injurious to a developing fetus. These hazards range from slippery floors to uncontrolled cigarette smoke, asbestos, known and unknown carcinogenic materials used in the workplace and a plethora of other hazards of modern society. The employer is, of course, free to protect itself from financially ruinous lawsuits by purchasing insurance and by maintaining the degree of care required by law. Moreover, an employer might be justified in dismissing an employee for failure to comply with safety procedures designed to protect the employee and his or her offspring. The focus, however, is on health and safety, not the possibility of litigation. Of course, litigation may coincide with failure to take reasonably necessary steps to protect an employee’s health or safety.

In sum, we believe that potential liability is too contingent and too broad a factor to amount to a ‘.‘business necessity.” Rather, under our formulation of business necessity, the defense in a fetal protection case is justified by a genuine desire to promote the health of employee offspring, not by self-interest. We do not mean to suggest that a trial court should attempt to determine whether an employer’s fetal protection program was motivated by self-interest or legitimate concern for employee offspring. Often, such a policy will be motivated by both. Our purpose in discussing the Hospital’s contention that avoidance of potential litigation can support a business necessity is to make clear that our extension of the business necessity defense beyond a strict relationship to job performance is based on a higher public policy than simply protecting employers from lawsuits.

16 . When we say that the employee’s case of disparate impact and the employer’s defense of business necessity apply “automatically,” we mean simply that in the trial of a case such as this no additional evidence need be introduced on these points, and the trial judge may proceed to the next issue — whether there were acceptable alternative policies, discussed below — without further analysis. Our discussion of disparate impact so far has been intended for theoretical clarity, to show how we got to the next step that requires proof.