POSNER, Circuit Judge.
In 1991 the Occupational Safety and Health Administration promulgated a rule on occupational exposure to bloodborne pathogens. 56 Fed.Reg. 64004, 57 Fed.Reg. 29206, 29 C.F.R. § 1910.1030. The rule is designed to protect health care workers from viruses, particularly those causing Hepatitis B and AIDS, that can be transmitted in the blood of patients. Promulgated after a protracted notice-and-comment rulemaking proceeding, the rule and its supporting reasons occupy 178 densely packed pages in the Federal Register. Most employers in the health care industry have accepted the rule, which in essence requires compliance with procedures for health care workers recommended by the Centers for Disease Control (since renamed the Centers for Disease Control and Prevention), the federal agency responsible for the control of contagious diseases. Many of these employers, indeed, had adopted the procedures as soon as the CDC recommended them. Three employer groups, however, challenge the rule — dentists, represented by the American Dental Association, and medical-personnel and home-health employers, both represented by the Home Health Services and Staffing Association. Medical-personnel firms supply health care workers on a temporary basis to hospitals and nursing homes, while home-health firms supply such workers to patients at home.
AIDS is caused by a virus (HIV) that can be transmitted, among other means, by introducing the blood of an infected person into the bloodstream of an uninfected one. If blood of a dental or medical patient who is HIV positive spatters on a health care worker’s skin where the skin is cut or abraded, or the worker accidentally sticks himself with a scalpel or hypodermic needle or other medical instrument on which there is fresh blood of an HIV carrier, the worker may become infected — with, so far as anyone knows, invariably fatal results. The AIDS virus is not, however, robust, and is not easily transmitted by the sorts of contact that patients usually have with health care workers. As of 1991, there had been only 24 confirmed cases of U.S. health care workers infected with the AIDS virus by patients since AIDS was first diagnosed in 1981.
Hepatitis B is a far more common disease than AIDS, though less scary, publicized, or stigmatized. The Hepatitis B virus (HBV) produces antibodies that fight the virus but at the same time destroy liver cells in which the virus has lodged. Although most infected persons recover uneventfully, about 1 percent die and about 6 to 10 percent of adult (and a much higher percentage of child) victims of Hepatitis B become carriers. The virus is much more virulent than the AIDS virus, and the introduction of a carrier’s blood into another person’s bloodstream is a particularly efficient means of transmission. Unlike the AIDS virus, which cannot survive exposure to air, HBV can survive on the surface of a piece of clothing or other material at room temperature for a week and can thus be spread by dirty laundry. Also unlike the AIDS virus, there is a vaccine against HBV, effective in 85 to 97 percent of healthy adults who receive it. Nonetheless, because of the greater virulence of HBV and the fact that many health care workers are not vaccinated, patient-communicated Hepatitis B kills about 200 health workers in the U.S. per year — roughly 100 times the number of such workers infected by patient-communicated HIV.
The precautions against infection of health care workers by the two viruses is similar, except that the vaccine against [825]*825HBV offers a protection that has no counterpart with regard to HIV, and contaminated laundry poses a danger of spreading HBV that also has no counterpart with regard to HIV. OSHA’s rule reflects the public-health philosophy of “universal precautions,” which means precautions against the blood of every patient, not just the blood of patients known or believed likely to be carriers of HBV or HIV. The precautions are various. They include engineering controls (such as requirements for the location of sinks), work practice controls (such as standards of care in handling contaminated sharp instruments, such as needles), requirements for personal protective equipment such as gloves, masks,' goggles, and gowns, requirements for housekeeping (covering such things as the cleaning of contaminated surfaces and laundry and the disposal of contaminated waste), reporting requirements, and provisions for medical care. The rule requires the employer to offer employees who are at risk of exposure to the blood of patients the Hepatitis B vaccine at the employer’s own expense, though it allows the employees to decline to be vaccinated. An employee who is involved in an “exposure incident,” such as being stuck with a contaminated needle, must be offered at the employer’s expense a confidential blood test for HBV and HIV; that is, only the employee is entitled to the result of the test.
In deciding to impose this extensive array of restrictions on the practice of medicine, nursing, and dentistry, OSHA did not (indeed is not authorized to) compare the benefits with the costs and impose the restrictions on finding that the former exceeded the latter. Instead it asked whether the restrictions would materially reduce a significant workplace risk to human health without imperiling the existence of, or threatening massive dislocation to, the health care industry. For this is the applicable legal standard. Occupational Safety & Health Act, § 6(b)(5), 29 U.S.C. § 655(b)(5); Industrial Union Dept., AFL-CIO v. American Petroleum Institute, 448 U.S. 607, 642-45, 655-56, 100 S.Ct. 2844, 2864-65, 2870-71, 65 L.Ed.2d 1010 (1980) (the “benzene” case) (plurality opinion); American Textile Mfrs. Institute, Inc. v. Donovan, 452 U.S. 490, 509-12, 530-36, 101 S.Ct. 2478, 2490-92, 2500-04, 69 L.Ed.2d 185 (1981) (the “cotton dust” case). The agency focused on HBV rather than on HIV because of the minute number of health care workers who have been infected by the latter virus. It estimated that the rule would eliminate between 113 and 129 annual deaths of health care workers from Hepatitis B, and a somewhat higher figure (187 to 197) if deaths of nonworkers infected by health-care workers who (but for the rule) would be carriers are factored in as well. (In making this additional calculation, OSHA expressed an uncharacteristic, but as it seems to us commendable, concern with the indirect effects of its rule. On the other hand it did not consider the reduction in medical care that might result from the rule’s effect in making the practice of medicine more costly — more on this shortly.) Most of these deaths would be avoided by the vaccine, but by no means all, because the vaccine is not a hundred percent effective and, more important, because many health care workers refuse to be vaccinated. Hence the other parts of the rule would have a positive effect even on Hepatitis B; and there is no vaccine (or cure) for AIDS.
OSHA’s evaluation of the effects of the rule, relying as it does on the undoubted expertise of the Centers for Disease Control, cannot seriously be faulted, at least by judges. Hence we cannot say that the rule, viewed as a whole, flunks the test of material reduction of a significant risk to workplace health. As for the impact on the health care industry, OSHA estimated the total cost of compliance with the rule at $813 million a year, clearly not enough to break the multi-hundred-billion-dollar healthcare industry. The rule’s implicit valuation of a life is high — about $4 million — but not so astronomical, certainly by regulatory standards, Cass R. Sunstein, After the Rights Revolution: Reconceiving the Regulatory State 239 (1990) (App. B), as to call the rationality of the rule seriously into question, especially when we consid[826]*826er that neither Hepatitis B nor AIDS is a disease of old people. These diseases are no respecters of youth; they cut off people in their working years, and thus in their prime, and it is natural to set a high value on the lost years. Nor is death the only consequence of these diseases. AIDS causes protracted pain and disability before death, and Hepatitis B causes pain and disability and often permanent liver damage, even when the patient “recovers.”
No doubt the agency’s $813 million estimate is an underestimate. It ignores time costs — more precisely, many or most time costs, for the statement accompanying the rule does contain a cost estimate for “work practices.” But apparently it is limited to “handwashing/glove change” and to the added time for using “safety syringes.” Those are not the only time costs, and the rest seem to have been left out. Individually slight, the time costs of suiting up with protective clothing, and of other preparatory activities required by the rule, could be cumulatively significant. Certain efficiency losses were also excluded, of which we give an example later. But the petitioners made no effort in the rulemaking proceeding to quantify these costs or to provide any basis for supposing them to be huge.
OSHA also exaggerated the number of lives likely to be saved by the rule by ignoring lives likely to be sacrificed by it, since the increased cost of medical care, to the extent passed on to consumers, will reduce the demand for medical care, and some people may lose their lives as a result. The agency’s consideration of the indirect costs of the rule is thus incomplete. Cf. Competitive Enterprise Institute v. NHTSA, 956 F.2d 321 (D.C.Cir.1992); International Union, UAW v. OSHA, 938 F.2d 1310, 1320 (D.C.Cir.1991). How many lives the rule is likely to sacrifice, however, we do not know; and again the petitioners make no effort to come up with á number. So while $4 million doubtless underestimates the agency’s implicit valuation of each life actually likely to be saved by the rule, we do not know how great the underestimate is and we cannot resolve our doubts against the agency. We add that the $4 million ignores the benefits to workers who will be spared illness — for remember that 99 times as many people get Hepatitis B as die from it.
As an original matter we might have been inclined to think that the regulation of the safety of the medical and dental workplace could be left largely to the market, that doctors, dentists, and other health care workers have a stronger incentive than the government to protect themselves from health hazards at reasonable cost, that their employees are compensated in their wages for what is after all a modest risk, and that health care workers who refuse to be vaccinated against Hepatitis B are knowingly assuming the risk and should be left to bear the consequences without government interference. But the occupational safety and health law is constructed on different premises that we are not free to question, and perhaps the infectious character of HIV and HBV warrants even on narrowly economic grounds more regulation than would be necessary in the case of a noncommunicable disease.
The petitioners, moreover, do not attempt to blast the entire rule out of the water. They do not argue that its net contribution to the safety of health workers is likely to be trivial, or its net costs crushing. They do not contend that there should be no regulation of bloodborne pathogens. They zero in on the particular features of the rule that pinch their clienteles and argue that the pinch is so tight that we should invalidate the rule insofar as it applies to the industries that they represent. We start with the dentists. Mainly they object that they have been lumped in with medical personnel, such as surgeons, who are far more exposed to patients’ blood than dental workers are and, in other respects as well, work under relevantly different conditions. For example, although the rule is entitled a rule about blood borne pathogens, it requires universal precautions in any dental procedure in which the patient’s saliva may drip or spray or splatter on the dental worker. But it is not quite true that OSHA treated all branches of the health care industry in a lump. It gave separate consideration to [827]*827every point raised before it by the dental association. It pointed out that the saliva of dental patients frequently contains blood — even in such routine procedures as having one’s teeth cleaned by a dental hygienist — and that it is possible, though far from certain, that even a small quantity of blood, diluted by saliva or some other fluid, can sometimes be infective. This was not some fantasy of OSHA. The Centers for Disease Control, while generally exempting saliva from the list of body fluids to which universal precautions should apply, recommended “special precautions” for dental workers exposed to saliva from patients. Did the CDC mean universal precautions? Its language is unclear, requiring interpretation. OSHA was entitled to adopt an interpretation that leaned “on the side of overprotection rather than underprotection.” Industrial Union Dept., AFL-CIO v. American Petroleum Institute, supra, 448 U.S. at 656, 100 S.Ct. at 2871.
What OSHA did not do was attempt to disaggregate the risk industry by industry. While it carefully disaggregated the costs of compliance, to see whether any industry within the health care sector would be imperiled by the rule, it did not attempt to determine separately the risk of HIV or HBV infection in dentistry, in home-health services, in thoracic surgery, in ophthalmology, and so forth. It did not attempt to determine the number of dental employees, say, who have contracted Hepatitis B from their patients; and it inflated the fact that only 1 of the 24 health care workers infected by a patient with AIDS in the U.S. was a dental employee by calling it a “significant percentage” of all occupational HIV infections. Well, 4.16 percent is a significant percentage, but not a meaningful one in this case, given the smallness of the sample. Citing cases like International Union, UAW v. OSHA, 988 F.2d at 1322, the dental association argues that the finding that the benzene opinion requires OSHA to make — “that the workplaces in question are not safe,” 448 U.S. at 642, 100 S.Ct. at 2864—entails a determination of the safety (or riskiness) of, at the least, each type of workplace.
OSHA cannot impose onerous requirements on an industry that does not pose substantial hazards to the safety or health of its workers merely because the industry is a part of some larger sector or grouping and the agency has decided to regulate at wholesale. That would be an irrational way to proceed. But neither is the agency required to proceed workplace by workplace, which in the case of bloodborne pathogens would require it to promulgate hundreds of thousands of separate rules. It is not our business to pick the happy medium between these extremes. It is OSHA's business. If it provides a rational explanation for its choice, we are bound. Associated Builders & Contractors, Inc. v. Brock, 862 F.2d 63, 68 (3d Cir.1988). It explained that while the cost of compliance with the precautions that the CDC has recommended (and OSHA has required) against bloodborne pathogens varies in a readily determinable fashion from industry to industry, the risk of infection does not. The risk goes with practices (so protective clothing is required only where being splashed with blood or other infective liquid can-reasonably be anticipated, whether it is a dentist’s office or a hospital operating room) rather than with industries, and the rule is therefore based on practices rather than on industries. The HIV or HBV carrier bears menace with him as he makes the rounds from health care, provider to health care provider. The risk of blood splatters and needlesticks is greater in some medical procedures than in others, but a dental hygienist is as likely to be splattered by blood contained in saliva as js many a worker in a hospital or a doctor’s office. The idea behind requiring universal precautions for health care workers is to protect those workers in any situation in which there .is a nontrivial risk of physical contact with a patient’s blood, and these situations arise in dentists’ offices as well as in doctors’ offices and hospitals. OSHA was required neither to quantify the risk to workers’ health nor to establish the existence of significant risk to a scientific certainty. Industrial Union Dept., AFL-CIO v. American Petroleum Institute, supra, 448 U.S. at 655-56, 100 S.Ct. at 2870-71. [828]*828It is true that because fewer people have dental than medical insurance, and therefore more people pay for dental care out of their own pockets, the higher price of dentistry that is a likely consequence of the rule will have a greater impact on demand; and inadequate dental care is a source of pain and suffering. But again the dental association made no effort to quantify this impact, though techniques for doing so exist in economics.
As to the specific precautions required by the rule, the association makes a number of arguments, of which a representative example is that children may be traumatized by the sight of a dentist wearing goggles. “Traumatized” is putting it too strongly, and the rule does not require goggles — at least for most dental procedures, ordinary glasses with side shields are all that is required. Nonetheless the problem of calming children in the dentist’s chair, without resort to nitrous oxide, is a serious one that the rule may aggravate, but again the dental association makes no effort to estimate the gravity of the harm. And while as we have suggested there is a time cost to decking oneself out in protective clothing which OSHA ignored, so does the dental association. In this example, and others unnecessary to discuss, the association is contesting requirements that, whether wise or not, are within the broad bounds of the reasonable, involving as they do technical issues on which the judgments of the CDC and OSHA are entitled to respect by the nonspecialist, biomedically unsophisticated Article III judiciary, at least in the absence of a more systematic showing of harms than attempted by the dental association.
This is true even with regard to the question whether a health worker should be permitted to conceal his awareness of being infected from his employer and the employer’s patients. The problem with nonconfidential medical tests is that the cost of flunking is high — loss of a job, for example — and this deters people from taking the test. A dental worker who knew that if he were infected he would lose his job might simply not report an exposure incident. The dental association does not argue that testing should not be encouraged, or alternatively that all dental workers should be tested regularly without regard to any exposure incidents. It asks us to balance the pros and cons of the confidentiality provision. That is not our job. We add that nothing in the rule forbids a dentist to require his employees to have a periodic HIV or HBV test, with disclosure of the results to him. The rule’s provision on confidentiality is limited to the test that an employer is required to offer a worker after an exposure incident.
The dental association complains that the rule goes too far in requiring dentists to “ensure” that their employees comply with the requirements of the rule. They say this imposes strict liability, which OSHA acknowledges it cannot do. Brennan v. OSHRC, 502 F.2d 946, 951 (3d Cir.1974); Brennan v. Butler Lime & Cement Co., 520 F.2d 1011, 1017 (7th Cir.1975); Pennsylvania Power & Light Co. v. OSHRC, 737 F.2d 350, 354 (3d Cir.1984). In so saying they may seem not really to be challenging the rule but rather to be raising an interpretive question — how strictly will OSHA interpret “ensure”? Interpretive questions that cannot be answered until a rule is applied are premature when raised in a challenge to the rule on its face, mounted at the time the rule is promulgated. But here the question is neither unanswerable, nor severable from questions that clearly are ripe, such as whether the costs of compliance with the rule will be so staggering as to imperil the dental industry (in which event the rule would be invalid). The stricter the liability, the more costly the rule. It is reasonably plain, however, that OSHA did not by using the word “ensure” seek to impose strict liability. It explained that the employer’s responsibility doesn’t end with furnishing his employees with protective gear, for example; he must do everything he can reasonably be expected to do to see that they use it. Like an employer made liable for his employees’ conduct not by the principle of respondeat superior (a form of strict liability) but by the negligence principle, Lancaster v. Norfolk & Western Ry., 773 F.2d 807, 818-19 [829]*829(7th Cir.1985), which requires due care in hiring, training, supervising, monitoring, disciplining, and retaining employees — the kind of employer liability imposed in sexual harassment cases, Guess v. Bethlehem Steel Corp., 913 F.2d 463, 465 (7th Cir.1990), in other civil rights cases, Hunter v. Allis-Chalmers Corp., 797 F.2d 1417, 1421-22 (7th Cir.1986), and in cases under the common law’s “fellow servant” doctrine, Pomer v. Schoolman, 875 F.2d 1262, 1266 (7th Cir.1989)—the employer subject to OSHA’s rule on bloodborne pathogens must take all reasonable measures to prevent his employees from violating the rule, but if despite these measures the employee violates the rule, the employer is off the hook. Pennsylvania Power & Light Co. v. OSHRC, supra; Capital Electric Line Builders of Kansas, Inc. v. Marshall, 678 F.2d 128 (10th Cir.1982); Brock v. L.E. Myers Co., 818 F.2d 1270, 1277 (6th Cir.1987).
The costs of compliance with OSHA’s rule, once the issue of strict liability for unforeseeable misconduct by employees is laid to one side, can hardly be thought so great as to imperil dentistry. Annualized, these costs are estimated to be equal to less than one-third of one percent of the industry’s annual revenues. This may overstate the actual cost, not to society as a whole (International Union, UAW v. OSHA, supra, 938 F.2d at 1320) but to the industry. When an industry is subjected to a higher cost, it does not simply swallow it; it raises its price and reduces its output, and in this way shifts a part of the cost to its consumers and a part to its suppliers (granted, those suppliers may include the firms constituting the industry). This very point is the basis of the dental association’s argument that OSHA’s rule is likely to cause a deterioration in dental care as dental patients flee the higher prices resulting from the industry’s efforts to shift some of the costs of compliance with the rule to its customers. There are some omitted costs, as we have noted, but not enough to make a decisive difference; nor does the association emphasize them.
The dental association makes some other jabs at the rule, but they have less merit than those we have discussed so we move on to the objections of the home health and medical personnel industries. These objections have greater force because they are based primarily on the fact, which has no counterpart in dentistry or for that matter in most other branches of the health industry, that the home health and medical personnel industries — we’ll call them the health personnel industry for short — do not control the sites at which their employees work. This does not affect compliance with the parts of the rule relating to HBV vaccination, post-exposure testing and treatment, and recordkeeping, but it does affect the ability of the employer to comply with the requirements for protective clothing and equipment, because his employees do not work in his presence. It especially affects the employer’s ability to comply with site-specific precautions required by the rule, such as ensuring that the work site is maintained in a clean and sanitary condition and that the worker has convenient access to running water in the event of exposure to blood or other potentially infective materials. These problems do not seem serious with regard to the branch of the industry that supplies medical personnel to hospitals, nursing homes, and other facilities that are themselves required to comply with the rule, but they could be serious with regard to the branch of the industry that supplies personnel to homes. Save for allowing the employer to substitute handwashing chemicals for a ready source of running water, the rule gives no recognition to the special problems of the industry although they were brought to OSHA’s attention during the rulemaking proceeding. OSHA’s brief relies on what it calls the “multi-employer worksite defense,” whereby an employer that cannot control a hazard is not liable for the exposure of its employees to the hazard if it took whatever precautionary steps were reasonable in the circumstances (or reasonably lacked the knowledge to recognize a condition as hazardous, but that is not a factor here). In effect OSHA asks us to read the defense into the rule.
[830]*830Is the absence from the rule of any explicit recognition of the acute compliance problem apparently faced by the health personnel industry, and in particular the branch that supplies medical workers to patients’ homes, a fatal omission? Administrative rules and decisions, like statutes, are enacted against a background of existing laws and understandings that do not have to be repeated in every new rule in order to have force. Perhaps the multi-employer worksite defense is so well established in the law of occupational safety and health that it formed part of the implicit background of the bloodborne-pathogens rule, which would explain OSHA’s seeming insouciance in assimilating the home personnel industry to health care industries that control the sites at which the care is provided. Apart from its having been clearly stated by the Occupational Safety and Health Review Commission, which reviews orders of OSHA citing employers for violations of the Occupational Safety and Health Act, and endorsed by every court to consider it, see, e.g., Anning-Johnson Co., 4 O.S.H.C. 1193, 1198-99 (1976); Harvey Workover, Inc., 7 O.S.H.C. 1687, 1689 (1976); D. Harris Masonry Contracting, Inc. v. Dole, 876 F.2d 343 (3d Cir.1989); see also Anning-Johnson Co. v. OSHRC, 516 F.2d 1081, 1089 (7th Cir.1975), the rule or some variant of it is implicit in the principle we have just examined that the Act does not impose strict liability.
One problem, however, is that the Review Commission is not OSHA. OSHA is legislator and prosecutor, OSHRC the judge. Martin v. OSHRC, — U.S. -, -, 111 S.Ct. 1171, 1174, 113 L.Ed.2d 117 (1991); see also Cuyahoga Valley Ry. v. United Transportation Union, 474 U.S. 3, 7, 106 S.Ct. 286, 288, 88 L.Ed.2d 2 (1985) (per curiam). By failing to mention the multi-employer worksite defense in the blood-borne-pathogens rule, OSHA may, despite its appellate lawyers’ disclaimer—which the doctrine of SEC v. Chenery Corp., 318 U.S. 80, 63 S.Ct. 454, 87 L.Ed. 626 (1943), obliges us to ignore — be signifying its disagreement that the defense is part of the law. If so, this could mean that OSHA was planning to enforce the rule as written, that is, without any such defense. If that is the proper interpretation of the rule’s silence, we would have to invalidate the rule (in part), as we think the defense is part of the law. The alternative interpretation of the rule’s silence is that OSHA accepts the defense as something that is to be read into all of its rules; that OSHA has bowed to the Review Commission and the courts.
We need not choose between these interpretations. Silence as acquiescence would not be an adequate response by OSHA to the circumstances of this case. The multi-employer worksite defense was developed for the construction industry and has, so far as we are able to discover, rarely been applied outside of it. What contours it should have in so unusual a worksite setting as the health personnel industry presents should not be left to appellate tribunals, whether OSHRC or this court, to decide. OSHA should set the contours in the first instance. Maybe they should be narrower in this industry than in the construction industry, maybe broader, maybe different. So important a question should be settled now, not left to enforcement proceedings. Since we know that the Occupational Safety and Health Act does not impose strict liability on employers, we know that the members of the health personnel industry need make only reasonable efforts to comply with the rule; but what this means at sites that the' employers do not control eludes us. Can the employer take the position that he has no responsibility for the condition of those sites? Or must he inspect each site before he allows his employees to work there? Must he do that even if the site is a private home? Who knows? OSHA had an obligation to consider such questions and the general issue that they present before imposing a medley of restrictions that, so far as appears, the industry cannot comply with.
So the rule must be vacated insofar as it applies to sites not controlled either by the employer or by a hospital, nursing home, or other entity that is itself subject to the bloodborne-pathogens rule. The other objections lodged by the health personnel in[831]*831dustry against the rule, however, either duplicate those of the dental association or plainly lack merit. So in the main the rule must be upheld. Which is not to say that it is a good rule. It may be unnecessary; it may go too far; its costs may exceed its benefits. Concern with the cost of health care in the United States is growing, and OSHA has received a steady drumbeat of criticisms even from supporters of public regulation of occupational health and safety. E.g., Sunstein, supra (index references to Occupational Safety and Health Act and Occupational Safety and Health Administration). But our duty as a reviewing court of generalist judges is merely to patrol the boundary of reasonableness, and, with the exception we have noted, OSHA’s blood-borne-pathogens rule — accepted as it has been by most health care industries and based as it is on the recommendations of the nation’s, perhaps the world’s, leading repository of knowledge about the control of infectious diseases — does not cross it.
The petition to review filed by the American Dental Association is denied. The petition of the Home Health Services and Staffing Association is granted in part and denied in part, as explained above.