No. 96-1643.United States Court of Appeals, First Circuit.Heard November 8, 1996
March 5, 1997.
[EDITORS’ NOTE: THIS PAGE CONTAINS HEADNOTES. HEADNOTES ARE NOT AN OFFICIAL PRODUCT OF THE COURT, THEREFORE THEY ARE NOT DISPLAYED.]
[EDITORS’ NOTE: THIS PAGE CONTAINS HEADNOTES. HEADNOTES ARE NOT AN OFFICIAL PRODUCT OF THE COURT, THEREFORE THEY ARE NOT DISPLAYED.]
John W. McCarthy, Bangor, ME, with whom Brent A. Singer and Rudman
Winchell were on brief, for defendant.
Peter M. Sfikas, Mark S. Rubin and Jill A. Wolowitz, Chicago, IL, on brief for American Dental Ass’n, amicus curiae.
Bennett H. Klein and David g. Webbert, Augusta, ME, with whom Gay and Lesbian Advocates and Johnson, Webbert Laubenstein were on brief, for plaintiff Sidney Abbott.
John E. Carnes, Commission Counsel, Augusta, ME, on brief for intervenor-plaintiff Maine Human Rights Commission.
Samuel R. Bagenstos, Attorney, U.S. Dept. of Justice, Washington, DC, with whom Deval L. Patrick, Assistant Attorney General, Boston, MA, and Jessica Dunsay Silver, Attorney,, Washington, DC, were on brief for intervenor-plaintiff United States of America, amicus curiae.
Robert Greenwald, Yonkers, NY, and AIDS Action Committee on brief, Boston, MA, for R.I. Dept. of Pub. Health, Bureau of Health of the Me. Dept. of Human Servs., American Pub. Health Ass’n, Ass’n of State and Territorial Dental Dirs., and Ass’n of State and Territorial Health Officials, amici curiae.
Donna Levin, Boston, MA, on brief for Mass. Dept. of Pub. Health, amicus curiae.
Appeal from the United States District Court for the District of Maine [Hon. Morton A. Brody, U.S. District Judge]
Before Selya, Circuit Judge, Aldrich, Senior Circuit Judge, and Stahl, Circuit Judge.
SELYA, Circuit Judge.
 In this case of first impression, the district court granted summary judgment in favor of plaintiff-appellee Sidney Abbott, an HIV-positive woman, on her claim that defendant-appellant Randon Bragdon, a dentist, violated the Americans with Disabilities Act (the ADA) by refusing to treat her in his dental office. Dr. Bragdon appeals. Because we agree with the district court that Ms. Abbott is disabled within the purview of the ADA and that providing routine dental care to her (i.e., filling a cavity) would not have posed a direct threat to Dr. Bragdon’s health or safety, we affirm.
 I. BACKGROUND
 The events giving rise to this litigation are straightforward. On September 16, 1994, Ms. Abbott arrived at Dr. Bragdon’s office in Bangor, Maine for a scheduled appointment. On her patient registration form, she indicated that she was infected with the HIV virus. People may be HIV-positive for years without manifesting the set of symptoms commonly known as AIDS, and Ms. Abbott was asymptomatic at the time.
 Dr. Bragdon performed a dental examination and discovered a cavity. He told Ms. Abbott that, pursuant to his infectious disease policy, he would not fill her cavity in his office, but would only treat her in a hospital setting. Though he would charge his regular fee, she would have to bear the additional cost of whatever the hospital charged for the use of its facilities. Ms. Abbott refused the
offer and instead filed a complaint under the ADA. See 42 U.S.C. §(s) 12182(a) (1994).
 After pretrial discovery concluded, the parties cross-moved for summary judgment. The district court ruled that Ms. Abbott was substantially limited in a major life activity (reproduction) and thus was disabled for purposes of the ADA. See Abbott v. Bragdon, 912 F. Supp. 580, 587 (D. Me. 1995). The court then concluded that the relatively routine treatment needed by Ms. Abbott could be delivered safely in Dr. Bragdon’s office. See id. at 591. Consequently, the court granted Ms. Abbott’s motion for summary judgment. See id. at 595-96. This appeal followed.
 II. THE SUMMARY JUDGMENT STANDARD
 The Civil Rules authorize federal courts to grant summary judgment only when “there is no genuine issue as to any material fact and . . . the moving party is entitled to judgment as a matter of law.” Fed.R.Civ.P. 56(c). Confronted with a properly documented motion for brevis disposition, the non-movant must establish the existence of a fact that is both genuine and material in order to ward off the entry of an adverse judgment. See Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 248 (1986); Garside v. Osco Drug, Inc., 895 F.2d 46, 48 (1st Cir. 1990). Appellate review of the district court’s award of summary judgment is plenary, and, in keeping with this standard, we are not wedded to the district court’s rationale but may affirm on any alternative ground made manifest by the record. See Hachikian v. FDIC, 96 F.3d 502, 504 (1st Cir. 1996).
 In assaying the record, we are guided by the same tenets that guided the lower court. Thus, we are duty bound to indulge all reasonable inferences in favor of the party opposing summary judgment. See id. This generous outlook notwithstanding, we must disregard improbable or overly attenuated inferences, unsupported conclusions, and rank speculation. See Smith v. F.W. Morse Co., 76 F.3d 413, 428 (1st Cir. 1996); Medina-Munoz v. R.J. Reynolds Tobacco Co., 896 F.2d 5, 8 (1st Cir. 1990).
 III. THE QUESTION OF DISABILITY
 The ADA sends a clear message to those who operate places of public accommodation: you may not discriminate against individuals in the full and equal enjoyment of services on the basis of a disability. See 42 U.S.C. §(s) 12182(a). Although a dental office qualifies as a place of public accommodation, see id. Section(s) 12181(7)(F); see also 28 CFR 36.104 (1996), the ADA protects only disabled patients against discrimination, and any attempt to invoke the ADA against a practicing dentist must start with an investigation into the patient’s status. We turn, then, to the question of whether Ms. Abbott, who was infected with HIV but was asymptomatic, had a disability cognizable under the ADA.
 This question is first and foremost a question of statutory construction which we review de novo. See Strickland v. Commissioner, Me. Dept. of Human Servs., 96 F.3d 542, 545 (1st Cir. 1996). In all such cases, we begin with the words of the statute, and we approach them with an understanding that our role is not to set public policy, but, rather, to discern the legislature’s will. See, e.g., United States v. Gibbens, 25 F.3d 28, 33 (1st Cir. 1994).
 A. The Plaintiff’s Burden.
 Disability is not a unitary concept under the ADA. Instead, the statute limns three subsets of disability, any one of which is sufficient to trigger the act’s protections. In this regard, the ADA states:
The term “disability” means, with respect to an individual — (A) a physical or mental impairment that substantially limits one or more of the major life activities of such individual; (B) a record of such an impairment; or (C) being regarded as having such an impairment.
 42 U.S.C. §(s) 12102(2). This case, as Ms. Abbott postures it, implicates the first subset of the statutory definition. Thus, she must
prove three things: first, that she has a “physical or mental impairment”; second, that this impairment adversely affects “a major life activity”; and third, that it does so to a significant extent (or, put more precisely, that the impairment “substantially limits” her ability to engage in the particular major life activity).
 B. The Impairment.
 The plaintiff easily clears the first hurdle. We hold unhesitatingly that HIV-positive status, simpliciter, whether symptomatic or asymptomatic, comprises a physical impairment under the ADA. Regulations issued by the Equal Employment Opportunity Commission (the EEOC) implementing Title III of the ADA explicitly support this conclusion, see 28 CFR Section(s) 36.104 (1996) (stating that the phrase “physical impairment” includes HIV); judicial authority buttresses this conclusion, see, e.g., Gates v. Rowland, 39 F.3d 1439, 1446 (9th Cir. 1994); Doe v. Garrett, 903 F.2d 1455, 1459 (11th Cir. 1990), cert. denied, 499 U.S. 904 (1991); and Dr. Bragdon does not seriously advocate an antithetical view.
 C. The Major Life Activity.
 Moving to the second hurdle, Ms. Abbott cites reproduction as her affected major life activity. The court below accepted this asseveration. See Abbott, 912 F. Supp. at 587. Dr. Bragdon’s rebuttal is twofold. In the first place, he disputes that reproduction properly can be characterized as a major life activity. In the second place, he asserts that even if reproduction so qualifies in general, there is nonetheless an unresolved issue as to whether it qualifies in Ms. Abbott’s particular case. We address each of these assertions.
 1. Reproduction Writ Large. The question of whether reproduction in large constitutes a major life activity under the ADA is not free from doubt. The ADA itself does not define the term “major life activities,” and the few available judicial precedents reveal divergent opinions. Compare Pacourek v. Inland Steel Co., 916 F. Supp. 797, 804 (N.D. Ill. 1996) (finding that reproduction is a major life activity) and Erickson v. Board of Govs. of State Colleges, 911 F. Supp. 316, 323 (N.D. Ill. 1995) (same) and Cain v. Hyatt, 734 F. Supp. 671, 679 (E.D. Pa. 1990) (same) with Krauel v. Iowa Methodist Med. Ctr., 95 F.3d 674, 677
(8th Cir. 1996) (holding that reproduction is not a major life activity) and Zatarain v. WDSU-TV, Inc., 881 F. Supp. 240, 243
(E.D. La. 1995) (same). Still, it is clear that Ms. Abbott’s HIV-positive status has a profound impact upon her ability to engage in intimate sexual activity, gestation, giving birth, childrearing, and nurturing familial relations. Our society has long recognized the fundamental importance of each element of this cluster of activities, and our jurisprudence reflects this bias. See, e.g., Stanley v. Illinois, 405 U.S. 645, 651 (1972) (terming the rights to conceive and raise children “essential,” “basic civil rights,” and rights that are “far more precious . . . than property rights”) (citations and internal quotation marks omitted). Viewed against this backdrop, we think it is highly likely that Congress accorded comparable importance to these activities when it authored the ADA.
 The statute’s text appears to bear out this intuition. Because the term “major life activities” is not defined in the enactment, we are obliged to construe it in accordance with its natural (that is, ordinary) meaning. See Bailey v. United States, 116 S. Ct. 501, 506 (1995); Smith v. United States, 508 U.S. 223, 228
(1993). The Court has looked to familiar dictionary definitions in similar situations. See, e.g., Bailey, 116 S.Ct. at 506; Smith, 508 U.S. at 229. Following that model here lends support to the classification of reproduction as a major life activity. The plain meaning of the word “major” denotes comparative importance. See, e.g., The American Heritage Dictionary of the English Language 1084 (3d ed. 1992) (listing
“greater than others in importance or rank” as the initial definition of “major”); Webster’s Ninth New Collegiate Dictionary 718 (1989) (defining “major” as “greater in dignity, rank, importance, or interest”). These definitions strongly suggest that the touchstone for determining an activity’s inclusion under the statutory rubric is its significance — and reproduction, which is both the source of all life and one of life’s most important activities, easily qualifies under that criterion.
 The origins of the ADA’s language reinforce this conclusion. Congress lifted the term “major life activities” from the Rehabilitation Act of 1973, which used it in defining an “individual with handicaps.” See 29 U.S.C. §(s) 706(8)(B) (1988). In that milieu, the term was accorded “a broad definition, one not limited to so-called `traditional handicaps.'” School Bd. of Nassau County v. Arline, 480 U.S. 273, 280 n. 5 (1987). In transplanting this combination of words from the soil of the Rehabilitation Act to that of the ADA, Congress specifically directed retention of the original meaning. See 42 U.S.C. §(s) 12201(a) (1994). Had Congress sought to confine the definition of disability narrowly, it surely would have written new, more restrictive language instead of borrowing a descriptive phrase notable for its breadth. See Doe v. Kohn Nast Graf, P.C., 862 F. Supp. 1310, 1320 (E.D. Pa. 1994). It would be wholly inconsistent with this history to hold that Congress did not envision reproduction as a major life activity.
 In addition to the language of the ADA and the historical antecedents of that language, we are guided by the regulations, which define “major life activities” to “mean functions such as caring for oneself, performing manual tasks, walking, seeing, hearing, speaking, breathing, learning, and working.”28 CFR Section(s) 36.104 (1996). As the regulation itself clearly indicates, this enumeration is not meant to be exclusive, and reproduction — one of the most natural of endeavors — fits comfortably within its sweep. Furthermore, the portion of the regulations which defines physical impairments to include physiological disorders affecting the reproductive system, 28 CFR Section(s) 36.104 (1996), militates in favor of the same outcome. From the scope of the latter regulation, we deduce that its drafters considered reproduction to be a major life activity — otherwise, including reproductive disorders among the regulation’s roster of physical impairments would not have made much sense. See Pacourek, 916 F. Supp. at 1404-05.
 The appellant resists this conclusion. The regulation itself includes no general adage to aid courts in determining what constitutes a major life activity, and he invites us to adopt a limiting principle which will preclude a finding that reproduction qualifies. In his view, major life activities do not embody lifestyle choices, or, as he puts it, “activities that many people decide never to do.” This proposition has a modicum of decisional support. See Krauel v. Iowa Methodist Med. Ctr., 915 F. Supp. 102, 106 n. 1 (S.D. Iowa 1995) (“Some people choose not to have children, but all people care for themselves, perform manual tasks, walk, see, hear, speak, breathe, learn, and work, unless a handicap or illness prevents them from doing so.”), aff’d, 95 F.3d 674 (8th Cir. 1996). In addition, courts have used other formulations en route to concluding that reproduction is unlike the activities listed in the regulation and, therefore, not a major life activity. See, e.g., Krauel, 95 F.3d at 677
(emphasizing that the plaintiff “has the ability to care for herself, perform manual tasks, walk, see, hear, speak, breathe, learn, and work” and is therefore not disabled); Zatarain, 881 F. Supp. at 243 (distinguishing reproduction from major life activities based on frequency of performance).
 We do not find any of these formulations persuasive. In Krauel, the Eighth Circuit did not go beyond the activities explicitly included in the regulation and thus effectively treated the list as exclusive, not illustrative.
Since the plain language of the regulation counsels otherwise, we are disinclined to emulate that example. The approaches taken by the Zatarain court and the district court in Krauel are no more attractive; in contradistinction to those courts, we see no reason why an activity must be performed either frequently or universally before it can be classified as a major life activity. There is no evidence that Congress intended either frequency or universality to operate as a restriction on the definition of “major life activities.” Indeed, the activities explicitly enumerated in the regulation are not wholly characterized by frequency and universality; learning — even in a broad sense — is for many adults not a part of daily life, and work is certainly not universal (as the lives of some of the very rich and some of the very poor demonstrate). The view from the other end of the spectrum bolsters this conclusion; neither lack of frequency nor lack of universality diminishes the fundamental importance of conceiving, childbearing, and raising a family.
 This leaves us with the bare bones of Dr. Bragdon’s thesis: that reproduction cannot be considered a major life activity because it is at bottom a lifestyle choice. That emaciated argument lacks force. To treat reproduction as a lifestyle choice, and no more, is merely an exercise in semantics. Most acts that human beings perform — or refrain from performing — have elements of volition. Speaking is undoubtedly a major life activity, but there are those (say, monks who have taken vows of silence) who choose to eschew it.
 Though the question is very close, we think it must be resolved favorably to Ms. Abbott. Reproduction (and the bundle of activities that it encompasses) constitutes a major life activity because of its singular importance to those who engage in it, both in terms of its significance in their lives and in terms of its relation to their day-to-day existence. Mindful of this reality, and honoring what we believe to be Congress’ intent, we hold that reproduction is a major life activity within the meaning of the ADA.
 2. Reproduction Writ Small. Dr. Bragdon’s fallback position is that Ms. Abbott is not disabled within the purview of the ADA unless reproduction is a major life activity for her. He then endeavors to use this position to defeat summary judgment, maintaining that a factfinder, drawing defendant-friendly inferences from the summary judgment record, would confront a genuine issue as to whether giving birth and raising a child is so important to Ms. Abbott as to constitute one of her major life activities.
 The premise on which Dr. Bragdon’s argument depends is of uncertain reliability. Though it is true that analysis under the first subset of the ADA’s definition of disability — “a physical or mental impairment that substantially limits one or more of the major life activities of [the plaintiff]” — calls for an individualized inquiry into whether the plaintiff is disabled, see 29 CFR Pt. 1630, App. 1630.2(j) (1996) (noting in the context of the ADA’s employment discrimination regulations that “[s]ome impairments may be disabling for particular individuals but not for others”); Katz v. City Metal Co., 87 F.3d 26, 32 (1st Cir. 1996); Ennis v. National Ass’n of Business Educ. Radio, Inc., 53 F.3d 55, 59 (4th Cir. 1995), the need for this case-by-case analysis of disability does not necessarily require a corresponding case-by-case inquiry into the connection between the plaintiff and the major life activity. By way of example, it might be enough for a court to consider only whether a given impairment substantially limits a particular plaintiff without considering whether the activity is of particular import to her. Be that as it may, the question is not dispositive here (nor will it be in many cases), and the most efficient way to decide this appeal is simply to postpone a definitive answer and assume, favorably to the appellant, that a plaintiff claiming a disability under the ADA must show a nexus between her impairment and the major life activity that she asserts has been substantially limited.
 Even accepting arguendo that reproduction is not a major life activity for one with no interest in bearing children, the determination of whether reproduction is a major life activity in a particular case does not automatically become a jury question. Cf. Medina-Munoz, 896 F.2d at 8 (noting that even “elusive concepts such as motive or intent” sometimes can be decided on summary
judgment). Here, drawing all reasonable inferences in the manner most helpful to Dr. Bragdon, a factfinder could reach no conclusion other than that reproduction, if a major life activity at all, constitutes such for Ms. Abbott. We explain briefly.
 Ms. Abbott’s testimony on this point is uncontradicted:
I have made the decision after I tested positive [for HIV] not to have children because of the risk of infecting the child and the risk of impairing my own immune system, and also the fact that this baby probably wouldn’t have a mother after a while.
 Dr. Bragdon offers no substantial rebuttal to Ms. Abbott’s assertion that HIV ended her consideration of having a family, but instead asks us to doubt her sincerity. In his most telling sortie, he stresses the fact that, during her deposition, Ms. Abbott replied “no” when asked, “Are you impaired in your ability to carry out any of your life functions by the fact that you are HIV positive?” We do not believe that a party’s response to a question about “life functions” — a phrase not used in ordinary parlance to refer to reproduction — can fairly be read as a direct contradiction of her unequivocal statements about her reasons for not bearing children. Dr. Bragdon has offered no other evidence suggesting that, short of her HIV infection, Ms. Abbott would have elected to remain childless. In the absence of specific discrediting evidence, a party cannot derail summary judgment by the primitive expedient of insisting that his opponent’s evidence should be disbelieved. See Grubb v. KMS Patriots, L.P., 88 F.3d 1, 4 (1st Cir. 1996).
 To say more at this point would be supererogatory. Because uncontradicted evidence establishes that reproduction is a major life activity for Ms. Abbott, the second element of her case is firmly in place.
 D. The Limitation.
 We turn now to the final hurdle that blocks Ms. Abbott’s path. At this hurdle, the parties joust over whether Ms. Abbott’s HIV infection substantially limits her major life activity of reproduction. Under the applicable regulation, a person’s major life activity is “substantially limited” if it is “restricted as to the conditions, manner, or duration under which [it] can be performed in comparison to most people.” 28 CFR Pt. 36, App. B Section(s) 36.104 (1996). Dr. Bragdon concedes that an HIV-positive pregnant woman faces an approximately 25% risk of transmitting the virus to her child without AZT therapy and an 8% risk of viral transmission with such therapy. He strives to persuade us, however, that there is an unresolved issue, sufficient to preclude summary judgment, as to whether HIV substantially limits Ms. Abbott’s reproductive activity. In other words, he claims that on the record sub judice, a jury reasonably could find that the stated degree of risk does not substantially limit this infected person’s ability to reproduce.
 We are unconvinced. No reasonable juror could conclude that an 8% risk of passing an incurable, debilitating, and inevitably fatal disease to one’s child is not a substantial restriction on reproductive activity. Cf., e.g., 29 CFR Pt. 1630, App. Section(s) 1630.2(j) (1996) (stating in the ADA’s employment discrimination regulations that “[o]ther impairments, however, such as HIV infection, are inherently substantially limiting”). In addition, Ms. Abbott faces the unfortunate reality that even if she gives birth to a healthy child, she probably will not live long enough to complete the task of raising the child to adulthood. We thus hold that HIV-positive status is a physical impairment that substantially limits a fecund woman’s major life activity of reproduction. Ms. Abbott therefore is disabled within the purview of the ADA.
 We add an eschatocol. We emphasize that the ADA’s “of such individual” language necessitates a factual determination based on the record; our discussion of this issue, therefore, focuses on the case before us. Each case presents different underpinnings, both factual and circumstantial. Presented with other facts and circumstances in a future case, perhaps reflecting dramatic improvements in medical science that substantially reduce the likelihood of transmitting HIV through reproduction, we might well reach a different conclusion than the one that we reach today. Such an ebb and flow is to be expected, because this is the very nature of the inquiry that the ADA mandates.
 IV. THE DIRECT THREAT QUESTION
 Under the ADA, a place of public accommodation must extend its services to all disabled persons without reference to their disabilities, subject to certain carefully circumscribed exceptions. See 42 U.S.C. §(s) 12182(a). By virtue of one such exception, a covered service provider need not deal with an individual who “poses a direct threat to the health or safety of others.” 42 U.S.C. §(s) 12182(b)(3). The term “direct threat” is defined by the statute; in this context it contemplates the existence of “a significant risk to the health or safety of others that cannot be eliminated by a modification of policies, practices, or procedures or by the provision of auxiliary aids or services.” Id.
 Dr. Bragdon seeks safe harbor under this exception, asserting that requiring him to treat Ms. Abbott would pose a direct threat to his health. His argument envisions that to fill Ms. Abbott’s cavity, he would need to inject a local anesthetic into her mouth and drill the decayed tooth. These procedures, he says, create an undue risk of HIV transmission both through needlestick and through the spattering of blood and bloody saliva.
 A court’s goal in conducting a direct threat analysis under the ADA is to achieve a responsible balance, protecting service providers and other places of accommodation from enforced exposure to unacceptable health and safety risks while at the same time protecting disabled individuals from discrimination that is rooted in prejudice or baseless fear. See Arline, 480 U.S. at 287. EEOC regulations drawn from the Court’s seminal opinion in Arline guide this analysis:
 In determining whether an individual poses a direct threat to the health or safety of others, a public accommodation must make an individualized assessment, based on reasonable judgment that relies on current medical knowledge or on the best available objective evidence, to ascertain: the nature, duration, and severity of the risk; the probability that the potential injury will actually occur; and whether reasonable modifications of policies, practices, or procedures will mitigate the risk.
 28 CFR Section(s) 36.208(c) (1996); see also Arline, 480 U.S. at 287-88 (discussing essentially the same factors).
 Of course, any inquiry along these lines depends in one sense on what information permissibly may be weighed in the balance and in another sense on the extent to which particular kinds of evidence should be assigned extra weight (or, perhaps, decretory significance). Thus, before evaluating the medical evidence proffered by the parties, we must determine (1) the relevancy, if any, of subsequent medical knowledge (that is, medical evidence not available in September 1994 when Dr. Bragdon refused to treat Ms. Abbott), and (2) the degree of deference, if any, due the judgments of public health officials.
 A. The Evidentiary Time Line.
 The first of these excursions need not detain us. The applicable regulations instruct that a judgment on the presence or absence of a direct threat must be predicated on “current medical knowledge or on the best available objective evidence,”28 CFR 36.208(c) (1996) (emphasis supplied), and, hence, point unwaveringly toward confining medical evidence to that available at the time a dentist or other health-care professional refuses to treat. This principle — that neither the service provider nor the prospective recipient of the service may prove or disprove the direct threat defense by relying on medical evidence not available when treatment was refused — is sound policy. Under such a regime, the service provider retains the opportunity to prove that he made an appropriate determination of the existence of a direct threat based on the evidence available when he made the decision to withhold his services. This is sound policy because, while health-care providers can be expected to maintain a working knowledge of currently accepted thinking in their fields, they cannot be expected to anticipate either future scientific advances or the emerging wisdom of public health organizations. In short, by holding covered service providers to an objective standard featuring the best evidence available at the time they refuse to render treatment to disabled persons, the ADA holds the delicate balance between sometimes conflicting rights steady and true.
 This approach is also scrupulously fair. To punish providers when they satisfy an objective standard based on the best evidence available at the time of their decisions would be to punish them for a lack of clairvoyance. By the same token, to hold providers harmless after they have refused treatment based on nothing more than unfounded trepidation would run at cross-purposes with the central theme of the ADA. Fundamental fairness insists that providers in such circumstances ought not to be entitled to rely on subsequent understandings to shield them from the condign consequences of discriminatory conduct.
Cf. McKennon v. Nashville Banner Pub. Co., 115 S.Ct. 879, 885 (1995) (holding that an employer in an employment discrimination case may not justify its conduct based on evidence that did not motivate it at the time of the employment decision); North Shore Univ. Hosp. v. Rosa, 657 N.E.2d 483, 486 (N.Y. 1995) (evaluating conduct alleged to be discriminatory under state law based on accepted medical practice at the time of the alleged infraction).
 B. The Degree of Deference.
 The second threshold determination involves the degree of deference due the medical judgments of public health authorities. The government joins Ms. Abbott in arguing for a rule which, if embraced, would cede great deference to those authorities. They posit that, in the absence of dissent among public health authorities, a service provider should be bound to accept the expressed collective judgment of those authorities unless he can demonstrate that this judgment is medically unreasonable. In contrast, Dr. Bragdon asseverates that, at least in the case of a service provider who is himself a skilled professional (such as a doctor or dentist), a court should defer to the provider’s judgment, as long as it appears to have been reasonable in light of then-current medical knowledge.
 The question of deference pivots on language in Arline, where Justice O’Connor, writing for the Court, stated that in making factual findings of the sort that are necessary to inform the inquiry into the existence vel non of a direct threat, “courts normally should defer to the reasonable medical judgments
of public health officials.” 480 U.S. at 288. In crafting regulations applicable to the ADA, the EEOC treated this passage from Arline as gospel. See 28 CFR Pt. 36, App. B 36.208 (1996) (remarking that the direct threat regulations “codif[y] the standard first applied by the Supreme Court in [Arline]”).
 We agree that the deference due public health officials must flow from the quoted passage — but acknowledging the hegemony of Arline does not signal automatic victory for Ms. Abbott and the government. The “defer entirely” formulation that they urge upon us is totally unprecedented; we have found no case in which the views of public health authorities are treated with the solicitude that Ms. Abbott and the government invite. If adopted, this formulation would come close to making a consensus among public health authorities unchallengeable by other medical evidence and, consequently, unreviewable by the courts. Nothing in Arline demands such obsequious obeisance to public health authorities or indicates an intention on the Court’s part to consign the medical judgments of private physicians to some evidentiary Siberia. Instead, the Court quite clearly left the details of deference for another day. See, e.g., Arline, 480 U.S. at 288 n. 15 (“This case does not present, and we do not address, the question whether courts should also defer to the reasonable medical judgments of private physicians . . . .”).
 Because we are unprepared to say that medical wisdom resides exclusively in public health authorities, we reject the idea of a conclusive presumption. The applicable regulations state that “[s]ources for medical knowledge include guidance from public health authorities, such as the U.S. Public Health Service, the Centers for Disease Control, and the National Institutes of Health.” 28 CFR Pt. 36, App. B Section(s) 36.208 (1996) (emphasis supplied). This list is plainly illustrative, not exhaustive, and the use of the verb “include” indicates to us that other sources of medical knowledge are within the pale. The statute, the suggestion implicit in the regulations, and the teachings of the Court are best synthesized by fashioning a rule which gives prima facie force to the views of public health authorities, but which permits a service provider to challenge those views based on contrary, properly supported opinions voiced by other recognized experts in the field (e.g., research studies published in peer-reviewed journals). Such a rule accords a meaningful degree of respect to the views of public health authorities, particularly when those views are unanimous. But the rule draws a distinction between respect and absolute capitulation. Under it, the conclusions of public health authorities may be rebutted by persuasive evidence adduced from other recognized experts in a given field.
 Treating the presumption of correctness which attaches to the collective judgment of public health authorities as rebuttable will not, as the government intimates, sabotage the statutory scheme. Because the test for the existence vel non of a direct threat remains an objective one, a service provider cannot successfully contradict an achieved consensus simply by proffering an unsupported opinion. This ensures that, despite the rebuttable nature of the presumption, the direct threat defense may not be used to mask prejudice or unfounded fears. Rather, to frame a genuine issue, an opposing view must be documented by competent countervailing evidence that is directly relevant. Speculative inferences, glancing statistics, unsupported conclusions, and ruminative surmise will not serve.
 C. Applying the Standard.
 We turn next to a review of the medical evidence that was available when Ms. Abbott visited Dr. Bragdon’s office in September 1994. By then, both the United States Centers for Disease Control and Prevention (CDC) and the American Dental Association (the Association) had spoken to the issue of the health risk to dental workers from patients infected with HIV. The Association’s 1991 Policy on AIDS, HIV Infection and the Practice of Dentistry stated that:
Current scientific and epidemiologic evidence indicates that there is little risk of transmission of infectious diseases through dental treatment if recommended infection control procedures are routinely followed. Patients with HIV infection may be safely treated in private dental offices when appropriate infection control procedures are employed.
 In 1993, the CDC updated its earlier guidelines and specified a compendium of infection control procedures, known as the “universal precautions,” for use by dental workers treating HIV-positive patients. See CDC, Recommended Infection-Control Practices for Dentistry, 1993 (the Recommendations). The CDC took the position that, when implemented, the prescribed precautions “should reduce the risk of disease transmission in the dental environment.” Id. at 3. While the guidelines do not state explicitly that no further risk-reduction measures are desirable or that routine dental care for HIV-positive individuals is safe, those two conclusions seem to be implicit in the guidelines’ detailed delineation of procedures for office treatment of HIV-positive patients. See United States v. Morvant, 898 F. Supp. 1157, 1166 (E.D. La. 1995) (concluding that “the universal precautions as prescribed by the CDC are universally accepted as `reasonable modifications’ of practices that will significantly mitigate the risk [of HIV transmission from patient to dentist]”). Tellingly, no public health authority has suggested that it is unsafe to provide routine dental care to HIV-positive patients in a private office environment. We find, therefore, that Ms. Abbott adduced competent evidence of reasonable medical judgments by public health officials, not contradicted by other public health authorities, to the effect that affording routine dental care (such as filling cavities) to HIV-infected patients in an office environment does not pose a direct threat to the dentist’s health.
 The next question is whether Dr. Bragdon has produced sufficient countervailing evidence that filling Ms. Abbott’s cavity in an office setting would have constituted a direct threat to his health. In an endeavor to create a genuine issue of material fact, he cites eight sources of information which he argues show that rendering the necessary treatment in his office would have jeopardized his health. We examine these sources to determine if any of them, individually or in the aggregate, justify denying summary judgment to Ms. Abbott.
 1. Dr. Bragdon notes that the Food and Drug Administration (the FDA) recommended in 1992 that persons who have had contact with a patient’s blood through needlestick, non-intact skin, or mucous membranes refrain from donating blood for a year. This recommendation is clearly insufficient to demonstrate a direct threat to Dr. Bragdon. To safeguard the integrity of the blood supply, the FDA may seek to avoid minute risks and take unusual precautions even in the absence of actual evidence of danger. It is not surprising, therefore, that the FDA issued its recommendation without making any finding that there was a significant risk of contracting HIV from contact of the type and kind described. In contrast, to determine whether there is a direct threat within the purview of the ADA, an inquiring court must gauge the nature, duration, and
severity of the risk. The FDA’s recommendation does not advance this inquiry.
 2. Dr. Bragdon refers to a CDC report documenting forty-two incidents of transmission of HIV to health-care workers and seven possible transmissions to dental workers. We deem this data insufficient to warrant depriving Ms. Abbott of summary judgment. Evidence of HIV transmission to health-care workers outside the dental field does not prove a direct threat to a practicing dentist in the absence of any evidence showing that the magnitude of risk to a dentist is comparable to the risk to other health-care workers in other settings. Nor is such an equivalency obvious; health-care workers in, say, emergency rooms may be exposed to much larger needles and more copious quantities of blood than are common in routine dental practice.
 Generalities about health-care workers aside, Dr. Bragdon does not cite a single confirmed instance of HIV transmission to a dentist. He does, of course, point to seven instances of “possible transmissions” of HIV to dental workers, but mere possibilities are too speculative to satisfy a litigant’s burden of production at the summary judgment stage. See Smith, 76 F.3d at 428 (noting that unsupported speculation must be disregarded at summary judgment); Medina-Munoz, 896 F.2d at 8 (similar).
 3. Dr. Bragdon brandishes a CDC report telling of the transmission of HIV from a Florida dentist to his patients. But neither the CDC report nor any other proffered evidence establishes transmission of HIV from a patient to a dental worker. The difference is meaningful. Dentists have the advantage of equipping and staffing their offices and dictating the precautionary procedures that will be utilized during patient encounters. In contrast, patients have no way of assuring that any particular risk-reduction measures, much less the universal precautions recommended by the CDC, are introduced. Given that dentists are in a superior position to patients in terms of protecting against infection, a single report of dentist-to-patient infection cannot reasonably be taken to suggest a direct threat to dentists from their patients’ infections.
 4. The high-speed drills that dentists use when filling cavities may also generate aerosol mists of water, blood, and bloody saliva. Using this datum as a springboard, Dr. Bragdon jumps to the conclusion that a study by researchers at the Stanford Medical School, raising concerns about transmission of HIV when dentists perform aerosol generating procedures, is sufficient to defeat summary judgment. See Gregory K. Johnson and William S. Robinson, Human Immunodeficiency Virus-1 (HIV-1) in the Vapors of Surgical Power Instruments, 33 Journal of Medical Virology 47 (1991). By its own characterization, however, the Johnson-Robinson paper is too conjectural to raise a genuine issue of material fact as to whether filling Ms. Abbott’s cavity would constitute a direct threat to Dr. Bragdon’s health. The paper notes that it “do[es] not quantitate the risk of HIV transmission . . . by such aerosols,” and that “the large body of epidemiologic data on prevalence of HIV infections . . . would suggest that transmission by aerosols is not common.” Id. at 49.
 5. Dr. Bragdon points out that the CDC did not state that it was medically unwise to take additional precautions with persons known to be HIV-positive. This observation, while true, gains him little ground. Such silence on the part of the CDC is at best equivocal; it does nothing to prove that there is any risk to a dentist in treating HIV-positive patients. Because the inference that Dr. Bragdon seeks to draw from the CDC’s silence is wholly conjectural, it cannot figure in the summary judgment calculus. See Smith, 76 F.3d at 428.
 6. Dr. Bragdon adverts to the Association’s report that the risk to health-care workers is greater than the risk to patients. At the same time, he notes, federal courts have found a significant risk of HIV transmission
from health-care workers to patients. See, e.g., Bradley v. University of Tex. M.D. Anderson Cancer Ctr., 3 F.3d 922, 924
(5th Cir. 1993) (per curiam), cert. denied, 510 U.S. 1119
(1994); Doe v. Washington Univ., 780 F. Supp. 628, 633 (E.D. Mo. 1991). Dr. Bragdon seeks to lace these findings together to support a conclusion that the risk to him from treating Ms. Abbott is also significant. The tie does not bind; this combination produces far too much of a generalization to have any force in the much narrower contours of this case.
 Moreover, the federal cases involving transmission of HIV from health-care workers to patients are inapposite not only because of the asymmetry of control of risk-reduction measures between health-care workers and patients, see supra, but also because the cited cases spring from a context in which any risk at all to patients is deemed unacceptable. See Washington Univ., 780 F. Supp. at 633 (“It is the stated goal of the medical profession to heal, and its secondary axiom, if healing is not possible, is not to harm.”) Therefore, these cases do not support a reasoned inference that rendering routine care to a dental patient infected with HIV constitutes a direct threat to the dentist’s health.
 7. Dr. Bragdon states that although he did his best to comply with the universal precautions, he still sustained sharp injuries on a regular basis. We believe that this kind of anecdotal evidence by a dentist who is not an expert on infectious disease is inadequate to block summary judgment. See Medina-Munoz, 896 F.2d at 8 (noting that summary judgment may be granted when opposing evidence is not significantly probative). Although courts need not defer slavishly to the judgments of public health officials, see supra Part IV(B), we believe it would be inconsistent with Arline were courts to credit lay testimony on matters of public health. See Arline, 480 U.S. at 287-88
(emphasizing the need to protect the handicapped from unfounded fears).
 8. Dr. Bragdon cites a study reporting that compliance with the universal precautions would reduce needlestick exposures by only 62%. See Edward S. Wong et al., Are Universal Precautions Effective in Reducing the Number of Occupational Exposures Among Health Care Workers?, 265 Journal of the American Medical Association 1123, 1126 (1991). This statistic says nothing about the initial baseline degree of danger of treatment in the absence of the universal precautions and thus is impuissant to prove that a dentist using the CDC’s precautions is directly threatened by treating an HIV-positive patient.
 At this point, we have reviewed all the proof relevant to direct threat that Dr. Bragdon claims was available to him in September of 1994. Each piece of evidence is too speculative or too tangential (or, in some instances, both) to create a genuine issue of material fact. This ends our item-by-item explication of the record.
 We next consider whether these eight proffers, in cumulation, possess greater probative force. This can occur when items of evidence, each insufficient in itself to prove a particular point, complement each other, like interlocking pieces of a jigsaw puzzle, in such a way that they together demonstrate some material fact. Thus, had Dr. Bragdon cited separate sources of evidence demonstrating (a) the likelihood of dental needlestick and (b) the likelihood of a dentist contracting HIV from a needlestick, these items together possibly would have been adequate to defeat summary judgment on the direct threat defense. Or, had he proffered several items of evidence that each showed a small risk to the health or safety of others, the aggregate effect of these items might have sufficed to prove a significant risk and thus to thwart summary judgment.
 Despite the fact that Dr. Bragdon did not explicitly make a “cumulative proof” argument, we have spontaneously reviewed the record with this thought in mind. Having done so, we are satisfied that the evidentiary proffers canvassed above, insufficient in themselves, are likewise insufficient in combination to call into legitimate question the lower court’s entry of summary judgment. In making this evaluation, we emphasize that, under the ADA, a service provider like Dr. Bragdon is not entitled to demand absolute safety; he can rely upon the direct threat defense only in response to significant risks. Here, Dr. Bragdon has failed to present
meaningfully probative evidence that treating Ms. Abbott would have posed a medically significant risk to his health or safety.
 V. CONCLUSION
 For the reasons indicated, we rule that Ms. Abbott’s HIV-positive status is a physical impairment which substantially interferes with her major life activity of reproduction, and that she is therefore disabled within the meaning of the ADA. Inasmuch as Dr. Bragdon has failed to produce sufficient evidence to establish a triable issue on his direct threat defense, the entry of summary judgment in Ms. Abbott’s favor must stand.
 Under ordinary circumstances, we would go no further. Here, however, we believe that more should be said. It is sometimes convenient to think of cases as involving conduct that may be categorized in terms of polar extremes: reasonable or unreasonable, praiseworthy or blameworthy, good or evil. But, given the complexities of the society in which we live, many decisions resist such facile classification into black-or-white dichotomies. Such cases are better characterized in varying shades of gray.
 This is such an instance. The litigants’ positions are understandable in human terms and impartial observers can empathize with both parties. Still, on the facts of record, the defendant’s refusal to render routine dental care to an HIV-positive patient offends a duly enacted federal statute and thus cannot be tolerated by a court of law.
 Although we do what we must, we are not blind to the difficulty of the choices that the ADA compels health-care professionals such as Dr. Bragdon to make. We also recognize that cases of this kind are necessarily fact-sensitive; had the patient required more invasive treatment or had the dentist proffered stronger evidence of a direct threat, the result may well have differed. We therefore caution future courts not to read our words more broadly than the context admits; our decision today eschews a blanket rule and instead demands case-by-case inquiry into a service provider’s responsibilities to treat HIV-positive patients.