C.R.S., a minor, by D.B.S., in his dual capacity as her father and natural guardian and as her guardian ad litem; D.B.S., in his capacity as an individual; N.A.S., in her capacity as an individual, Appellants, Minneapolis Painting Industry Health and Welfare Fund, Local 386, Intervenor, v. UNITED STATES of America, Appellee; C.R.S., a minor, by D.B.S., in his dual capacity as her father and natural guardian and as her guardian ad litem; D.B.S., in his capacity as an individual; N.A.S., in her capacity as an individual, Plaintiffs, Minneapolis Painting Industry Health and Welfare Fund, Local 386, Appellant, v. UNITED STATES of America, Appellee, 11 F.3d 791

MAGILL, Circuit Judge. D.B.S., N.A.S., and their minor child C.R.S. (plaintiffs) brought this suit under the Federal Tort Claims Act (FTCA), 28 U.S.C. §§ 1346(b), 2671-2680 (1988), alleging that they all contracted AIDS as a result of the federal government’s negligence in providing D.B.S. with contaminated blood when he underwent transfusions at the Martin Army Community Hospital (MACH), in Fort Benning, Georgia, while a member of the Minnesota National Guard. Plaintiffs appeal from the district court’s order granting defendant’s motion for

summary judgment. 820 F.Supp. 449. For the reasons set forth below, we affirm the district- court’s order. I. BACKGROUND In August 1983, D.B.S. was ordered to go to Fort Benning for basic training. While in training, D.B.S. required emergency medical care for an intestinal condition. He underwent surgery and blood transfusions at MACH in response to the condition. After recovering from his surgery, D.B.S. completed the training and returned to his home in Minnesota in December 1983. In November 1984, he married N.A.S. and they had three children, including C.R.S., who was born in June 1987.

In late 1987, the Minnesota National Guard discharged D.B.S. In early 1989, all three plaintiffs discovered that they had tested positive for the presence of the human immunodeficiency virus (HIV) after undergoing blood tests. In the summer of 1992, Donor 3903, one of nine donors whose blood MACH had used in D.B.S.’s transfusions, also tested HIV positive. At the time of D.B.S.’s transfusion in 1983, public health officials and blood banks were struggling to address the AIDS crisis. Researchers, for instance, had yet to develop the HIV test, so blood banks could not be certain whether donated blood was infected.

On January 4, 1983, the Centers for Disease Control organized a national meeting to discuss how blood banks could protect their supplies from contamination. While disputes arose at the meeting regarding what precautions would best protect the blood supply, attendees agreed that they should adopt some measures to keep intravenous drug users and homosexual and bisexual men from donating blood because these groups constituted nearly ninety percent of AIDS patients at the time.

In March 1983, the Food and Drug Administration (FDA) and the American Association of Blood Banks (AABB) issued recommendations concerning procedures for screening donors for AIDS. The recommendations suggested educating donors about AIDS, questioning them about the early signs of AIDS, and asking those in high risk groups not to give blood. Thus, the FDA/ AABB procedures relied on voluntary self-deferral by high risk donors and did not include measures such as direct questioning.

In a memorandum dated April 12, 1983, the Military Blood Program Office (MBPO) adopted the FDA/AABB guidelines, which were applicable to private sector blood banks. The MBPO is responsible for the operation of the military’s blood banks. Department of Defense Directive 6480.5 established the MBPO in 1972.

The directive stated that the MBPO was responsible for “[d]evelop[ing], recommend[ing] and monitor[ing] policies (fully coordinated with the Military Departments) on ... [t]he collection, procurement, processing, storage, distribution, and management of whole blood.” United States Department of Defense, Directive 6480.5, at 3 (June 16, 1972). It also directed the MBPO to “[m]aintain and disseminate plans for coordinating the collection, processing, distribution, and management of blood and blood components necessary to meet the requirements of the Military Departments.” Id.

The Assistant Secretary of Defense (Health and Environment), in contrast, was responsible for “providing overall policy guidance on the ... Military Blood Program, and for coordinating it with the National Emergency Blood Program, when required.” Id. at 4. Because the MBPO staff consisted of only three people, it often relied on FDA and AABB regulations in formulating military blood bank policies. Its reliance here is reflected in the April 12, 1983, memorandum adopting the FDA/AABB civilian guidelines. Indeed, Capt.

James Bates, the Director of the MBPO in 1983, testified that the MBPO did not consider adopting different procedures for the military. Thus, when Donor 3903 gave blood in August 1983, MACH’s blood bank had in place the procedures outlined in the MBPO memorandum, which included providing information about AIDS to potential donors, identifying high-risk groups, asking that members of these groups refrain from donating, and distributing cards prospective donors filled out that were updated to reflect concerns about AIDS.

The FDA/AABB guidelines did not call for notifying recipients of blood that they could have received infected blood. In 1985, however, the Department of Defense began testing active members of the military for HIV infection. Then, in February 1986, the Army Surgeon General circulated guidelines classifying those who received transfusions between January 1980 and July 1985 as being at high risk for HIV exposure. The guidelines then described ways to identify, notify, evaluate and counsel such individuals.

Plaintiffs brought suit in the United States District Court for the District of Minnesota in 1990 after four government agencies rejected their administrative claims. They asserted two negligence claims: (1) that the military was negligent in adopting the FDA/ AABB blood donor screening procedures in 1983, which they allege were unreasonable in the military context, and (2) that the military was negligent in later failing to warn D.B.S. of the risk that he had received contaminated blood at MACH. The district court denied defendant’s motion to dismiss or, in the alternative, for

summary judgment, that the action was barred by the Feres doctrine. The government then asserted the discretionary function exception to the FTCA, 28 U.S.C. § 2680(a) (1988), barred the claims. The district court granted this motion, finding that § 2680(

a) barred all claims. Plaintiffs then filed this appeal.

Summary judgment is appropriate when the record shows that “there is no genuine issue as to any material fact-and that the moving party is entitled to judgment as a matter of law.” Fed.R.Civ.P. 56(c). There is no genuine issue for trial “[w]here the record taken as a whole could not lead a rational trier of fact to find for the. non-moving party.” Matsushita Elec. Indus. Co. v. Zenith Radio Corp., 475 U.S. 574, 587, 106’ S.Ct. 1348, 1356, 89 L.Ed.2d 538 (1986). This court reviews the granting of .a motion for

summary judgment de novo. United States ex rel. Glass v. Medtronic, Inc., 957 F.2d 605, 607 (8th Cir.1992). In light of these standards, we first discuss the discretionary function exception and then' apply this law to each of plaintiffs’ claims. II. DISCUSSION Congress waived the sovereign immunity of the United States by enacting the FTCA, under which the federal government is liable for certain torts its agents commit in the course of their employment. See 28 U.S.C. § 2674 (1988).

The FTCA, however, does not provide a blanket waiver, but rather retains the government’s immunity from suit through several exceptions. One such exception is the discretionary function exception, which prohibits suits “based upon the exercise or performance or the failure to exercise or perform a discretionary function or duty on the part of a federal agency or an employee of the Government, whether or not the discretion involved be abused.” Id. § 2680(a). The Supreme Court has announced a two-part test for determining when the discretionary function exception applies.

The first requirement is that the conduct at issue indeed be discretionary, that is, it must “involve[ ] an element of judgment or choice.” Berkovitz v. United States, 486 U.S. 531, 536, 108 S.Ct. 1954, 1958, 100 L.Ed.2d 531 (1988). Agency decisions exercising delegated authority from Congress to establish programs and policies in implementing the general provisions of a regulatory statute, for example, involve one type of judgment that the exception protects. See United States v. Gaubert, 499 U.S. 315, 323-24, 111 S.Ct. 1267, 1274, 113 L.Ed.2d 335 (1991).

Decisions made at the operational leVel, as well as decisions made at the policy-planning level, can involve the exercise of protected discretion. See id. at 325-26, 111 S.Ct. at 1275. Moreover, “the fact that determinations are made at a relatively low level does not prevent the applicability of the exception.” Layton v. United States, 984 F.2d 1496, 1500 (8th Cir.) (citing Gaubert, 499 U.S. at 332-33, 111 S.Ct. at 1279), cert. denied, — U.S. -, 114 S.Ct. 213, 126 L.Ed.2d 170 (1993).

Judgment or choice is absent “when a federal statute, regulation, or policy specifically prescribes a course of action for an employee to follow” because then “the employee has no rightful option but to adhere to the directive.” Berkovitz, 486 U.S. at 536, 108 S.Ct. at 1958-59. The exception’s second requirement is that the judgment at issue be “of the kind that the discretionary function exception was designed to shield.” Id.

Because the exception’s purpose is to prevent “judicial ‘second-guessing’ ” of government decisions based on public policy considerations, it protects only those judgments “grounded in social, economic, and political policy.” United States v. S.A. Empresa de Viacao Aerea Rio Grandense (Varig Airlines), 467 U.S. 797, 814, 104 S.Ct. 2755, 2765, 81 L.Ed.2d 660 (1984). A decision, however, need not involve conscious consideration of policy factors by the government agent. See Gaubert, 499 U.S. at 325-26, 111 S.Ct. at 1275.

The focus of the inquiry, rather, is “on the nature of the actions taken, and whether they are susceptible to policy analysis.” Id. A. Screening Claim Plaintiffs’ first claim is that the MBPO was negligent in its failure to screen blood donors adequately for HIV. They claim that the defendant’s decision to adopt the civilian screening procedures of the FDA and the AABB was unreasonable because of special circumstances prevailing in the military.

The FDA/AABB procedures, for instance, relied on donors in groups at high risk for AIDS voluntarily to defer from giving blood, a system which plaintiffs argue was unlikely to work in an environment like the military which punishes those who engage in high-risk behavior such as homosexual contact and drug abuse. The defendant, they allege, should have designed its own more stringent procedures to address the threat of infected blood in light of these special circumstances. In short, plaintiffs challenge only the government’s decision, not its implementation of this decision.

Applying the Supreme Court's two-part test, we find that the defendant’s decision to adopt civilian screening procedures was a, protected exercise of judgment within the discretionary function exception. First, the decision was discretionary because it involved the exercise of judgment and choice. Indeed, there are no mandatory regulations or policies directing the MBPO to adopt particular blood screening procedures. To be sure, Directive 6480.5 directed the MBPO to “develop ... policies” concerning blood “collection, procurement, processing, storage, distribution, and management.” Directive 6480.5, supra, at 3.

The directive, however, vested the MBPO with broad discretion to choose the type of screening procedures it deemed appropriate, leaving even the choice of how to go about making this decision to the MBPO. Plaintiffs have failed to identify any regulation or policy that “specifically prescribe[d] a course of action” for the defendant to follow in selecting screening procedures. We thus hold that the MBPO had discretion in determining what screening procedures to adopt.

Plaintiffs contend, however, that the MBPO did not have policy-making authority over blood screening and thus had no authority even to make the decision to adopt FDA/ AABB procedures in the first place. They claim that Directive 6480.5 reserved all policy-making authority to the Assistant Secretary of Defense (Health and Environment) and that the MBPO was responsible merely for coordination and making recommendations. The plain language of the directive undermines these arguments. The directive stated that the Assistant Secretary was “responsible for providing overall policy guidance on the ...

Military Blood Program, and for coordinating it with the National Emergency Blood Program, when required.” Id. at 4. That the Assistant Secretary had authority to provide “overall policy guidance” does not affect the. fact that the MBPO itself had authority to, among other functions, “develop.. .. policies” and “maintain .... plans” relating to donor screening procedures. Nor does the directive suggest that all decisions relating to the military blood program were required to be independently approved by the Assistant Secretary.

Rather, the MBPO and the Assistant Secretary shared authority over implementing donor screening and blood collection procedures. We thus reject plaintiffs’ claim that the MBPO lacked authority to make its discretionary decision. Moreover, the defendant’s decision regarding screening procedures meets the second part of the Supreme Court’s test because it was susceptible to a balancing of social, economic, and political policy factors.

How to screen those donating blood the military is to use in its hospitals implicates a host of complex policy issues, ranging from the need to keep costs in check given the budget constraints under which government operates to the need to ensure that the blood supply is safe and plentiful. Moreover, the complicated public health issues surrounding AIDS were compounded here because at the time the decision to adopt FDA/AABB screening procedures was made, the medical community still had imperfect knowledge regarding many aspects of the disease.

In short, the issue of what screening procedures to adopt is precisely the type of policy-bound decision that Congress intended to insulate from judicial scrutiny through the discretionary function exception. Plaintiffs assert that the decision as to screening procedures was not a question of public policy, but rather that it was a technical matter of professional judgment which § 2680(

a) does not protect. They point to the fact that experts on neither side cited public policy factors in evaluating defendant’s conduct as evidence that the decision merely involved exercising “technical, medical/administrative” judgment. The first response to this argument is that there is no bright line rule removing “professional” or “technical” judgments, whatever these may be, from the scope of the exception. The inquiry does not depend on labels, but rather centers on whether government conduct satisfies the Supreme Court’s two-pronged test.

Here, we have already explained that the conduct at issue was both discretionary and grounded in policy considerations. Moreover, defendant’s conduct here is of a markedly different type than the government’s conduct in Aslakson v. United States, 790 F.2d 688 (8th Cir.1986), which plaintiffs cite for support. In Aslakson, plaintiff brought an FTCA claim against the government after his son died as a result of contact between the mast of his sailboat and a federal power line. Id. at 690.

The government’s policy “clearly required it to elevate its power lines if safety considerations compelled such action.” Id. at 693. The government asserted that § 2680(

a) protected its determination that the lines were not a safety hazard. See id. at 692. We held that “[wjhere the challenged government activity involves safety considerations under an established policy rather than the balancing of competing public policy considerations,” the discretionary function exception does not apply. Id. at 693.

The obvious distinction between this ease and Aslakson is that here, the government conduct did involve “the balancing of competing public policy considerations.” In contrast, in Aslakson, while deciding what constituted a safety hazard “involve[d] some degree of judgment on the part of government officials,” such judgment only “require[d] the knowledge and professional expertise” of the officials but did not implicate policy concerns. Id. A more- subtle distinction is that in Aslakson, the government had a safety policy already in place that officials failed to follow.

Here, claims that defendant was negligent in failing to comply with a mandatory safety policy are not'.before this court. Plaintiffs next argue that the record lacks any suggestion that defendant consciously balanced the various policy factors or considered alternative procedures in reaching its decision and that therefore the decision was not one grounded in policy.

Even assuming the record fails to show that the defendant in fact considered public policy factors, Gaubert makes clear that the relevant inquiry is merely whether the conduct at issue is “susceptible to policy analysis.” Gaubert, 499 U.S. at 325-26, 111 S.Ct. at 1275 (emphasis added); see also Sea-Land Serv., Inc. v.

United States, 919 F.2d 888, 892 (3d Cir.1990) (finding that a court “need not examine the record for evidence of a conscious policy decision” because “[a]n oversight or a failure to weigh the relevant factors does not affect whether ... the conduct [is] susceptible to policy analysis”)] cert. denied, — U.S. -, 111 S.Ct. 2235, 114 L.Ed.2d 477 (1991); Kennewick Irrigation Dist. v. United States, 880 F.2d 1018, 1028 (9th Cir.1989) (rejecting the argument that the government had to show that it “considered [policy] factors and made a conscious decision on the basis of them”).

Here, the issue of what screening procedures to adopt is plainly susceptible to policy analysis, as explained above. Defendant could have considered a wide range of policy factors in making its decision; whether or not it actually did so is immaterial under Gaubert. Thus, defendant’s decision satisfies § 2680(a)’s requirement that the government conduct at issue be grounded in public policy considerations. Plaintiffs advance one other argument in support of their claim that defendant’s screening decision falls outside of the discretionary function exception.

They claim that the exception is inapplicable because the defendant’s decision here related merely to internal governmental operations rather than the core functions of the military, such as national defense. This court, however, has already rejected such a categorical distinction between governmental and internal operations. In Jurzec v. American Motors Corp., 856 F.2d 1116, 1117 (8th Cir.1988), plaintiff-sued the government after her husband died as a result of the rollover of a surplus jeep he had purchased from the Postal Service. She asserted that § 2680(

a) did not apply because she was suing the Postal Service in the Service’s capacity as a seller of jeeps, “a traditionally non-governmental role,” rather than in its capacity as a regulator. Id. at 1118. We rejected this claim, finding that “[t]his logic runs square against Supreme Court precedent stating that ‘it is the nature of the conduct, rather than the status of the actor, that governs whether the discretionary function exception applies in a given case.’ ” Id. (quoting Varig Airlines, 467 U.S. at 813, 104 S.Ct. at 2764), Moreover, McMichael v.

United States, 751 F.2d 303, 306 (8th Cir.1985), upon which plaintiffs rely, is inappo-site. Plaintiffs cite McMichael in support of their claim that whether the government conduct involves a regulatory purpose is a crucial factor in determining the applicability of § 2680(a). While it is true that the McMi-chael court noted that the absence of a regulatory purpose militated against applying the exception, see id. at 306, the basis for finding the government subject to suit was that the government conduct at issue was not discretionary. See id. at 307.

Jurzec itself suggested that the holding in McMichael depended on the fact that the government lacked discretion, not on the. absence of a regulatory purpose. See Jurzec, 856 F.2d at 1119-20. Thus, we need not. determine here whether the military’s provision of medical services to its personnel is an internal operation or a core governmental function. This inquiry is simply not relevant to the question of whether challenged governmental action is discretionary and whether it is susceptible to policy analysis. B.

Notification Claim Plaintiffs’ second claim is that the government was negligent in failing to warn D.B.S. of the risk that he had received contaminated blood during his 1983 transfusions. They claim that once the laboratory test for HIV became available in 1985, defendant should have alerted those who had previously undergone transfusions in military hospitals of the possibility they had become infected. Such a warning, they assert, could have prevented transmission of AIDS from D.B.S. to his wife and child. Our analysis of this claim parallels that which we used for the screening claim.

The issue again is whether the defendant is entitled to

summary judgment because the government’s conduct is protected by the discretionary function exception. Thus, the first question is whether the defendant’s failure to warn was discretionary, that is, whether the conduct involved judgment or choice that had not been removed by statute or regulation.

Plaintiffs assert that a document entitled “Guidelines for the Evaluation, Counseling, and Management of HTLV-II1/LAV-Infected Individuals and Their Contacts” (the Guidelines), issued by the Army Surgeon General in February 1986, created a duty on the part of defendant to warn recipients of untested blood such as D.B.S. Under Berko-vitz, an agency policy strips government employees of discretion when it “specifically prescribes a course of action for an employee to follow.” 486 U.S. at 636, 108 S.Ct. at 1958.

Indeed, to remove discretion, the policy must constitute a “specific mandatory directive.” Id. at 544, 108 S.Ct. at 1963. Cases from other circuits have applied this language from Berkovitz. In Kennewick Irrigation Disk, the Ninth Circuit held that a safety standard removes discretion “when it is embodied in a specific and mandatory regulation or statute which creates clear duties incumbent upon the governmental actors.” 880 F.2d at 1026 (third emphasis added). Similarly, in Dube v.

Pittsburgh Corning, 870 F.2d 790, 793 (1st Cir.1989), the First Circuit considered a Navy safety regulation reading: “Each individual concerned shall warn others whom he believes to be endangered by known hazards or by failure to observe safety precautions.” Accepting the government’s argument that this language was not sufficiently specific under Ber-kovitz, the court held that “the broad language of the regulation suggests that those charged with its implementation retained wide latitude regarding its execution. [It created] no ‘specific mandatory directive’ to warn....” 870 F.2d at 794 (quoting Berko-vitz, 486 U.S. at 544, 108 S.Ct. at 1963).

Thus, to remove discretion from government employees, a regulation must be mandatory and it must clearly and specifically define what the employees are supposed to do. As the Court did in Berkovitz, we must now scrutinize the government policy at issue to determine if it “specifically prescribed] a course of action” for agency employees to follow. The Guidelines first defined various groups at risk of developing infection. Included under the heading “High Risk Category” were individuals who “received blood transfusions prior to 1 July 1985 but since 1 January 1980.” Appellants’ App. at 200.

D.B.S., because he received a transfusion in 1983, was in this category. Thousands of other people, however, were also in this category. For the Army to notify those in the risk categories, Army employees needed to determine their identities. The system the Guidelines provided for identifying persons in the defined risk groups was subsection B, “Identification of Contacts and Persons at Risk” (subsection B).

Subsection B stated that “[contacts of HTLV-III-infected individuals and persons at risk of developing infection will be identified through the interview of HTLV-III-infected persons or through self-referral.” Id. at 201. The next sentence stated that “[tjhese-per-sons will be notified by local health authorities.” Id. (emphasis added). Thus, the prerequisite to receiving notification was being first identified as a “contact[ ]” or “person[ ] at risk.” We conclude that the Guidelines were not a specific mandatory directive removing discretion from defendant regarding whether to notify people like D.B.S.

Subsection B is subject to at least two

interpretations. First, it could have meant that there were two methods of identifying persons at risk: the interview of infected persons and self-identification by the persons at risk. It is unlikely that this is what it meant, however, because the notion that persons at risk, as opposed to contacts, would be identified through the interview of infected persons is nonsensical. For instance, the Guidelines listed intravenous drug users as a high risk group. See id. at 200.

Interviewing an infected intravenous drug user would certainly identify other intravenous drug users with whom he shared needles and his sexual partners, i.e., his contacts. Such an interview would not, however, identify the other intravenous drug users around the country with whom the interviewee did not share needles and who he did not know, i.e., persons at risk. Thus, subsection B more likely meant that the Army was to identify contacts through the interview of infected persons and that persons at risk were to identify themselves through self-referral.

Under this reading, because he does not claim that he identified himself through self-referral, D.B.S. was not entitled to receive notification under subsection B. Even if the Guidelines did contemplate identifying persons- at risk through the interview of infected persons, however, the Army still had discretion under subsection B’s plain language not to notify D.B.S. of the risk he had received infected blood. Indeed, Army employees had discretion not to take any steps to identify persons at risk or contacts.

These people were to be identified through interviews of infected persons “or through self-referral.” Id. at 201 (emphasis added). Because subsection B stated these alternatives in the disjunctive, the Army had discretion to simply wait for persons at risk and contacts to identify themselves through self-referral and then follow Guidelines procedures as to counseling and testing. Under this reading, Army employees did not have to take the affirmative step of interviewing infected persons pursuant to subsection B.

Thus, from the face of the policy, Army employees had discretion not to notify people like D.B.S. who did not self-refer. Finally, even if we ignore all of the ambiguities in the first sentence of subsection B, plaintiffs’ argument still fails the requirement that the government policy be specific and clear. The Guidelines did state, in mandatory language, that persons in risk groups “will be notified” and “will need to be interviewed.” Id. As explained above, the Guidelines also defined these risk groups, included in which were those who received blood transfusions between 1980 and 1986.

Within these broad parameters that the Guidelines established, however, every single detail of what would be a massive project was left to the discretion of Army employees. Employees had discretion regarding when to begin the notification process, when to end it, how to prioritize the order in which people received notification, whether only active duty members would receive notification or also reservists and anyone who had been in the Army since the beginning of the AIDS crisis, and how to undertake the monumental task of identifying the thousands of people in the risk groups in order to notify them.

In short, as was the case in Dube, the Guidelines’ language “suggests that those charged with [the pokey’s] implementation retained wide latitude regarding its execution.” 870 F.2d at 794. Although they contained “mandatory” language, the Guidelines had broad and confusing provisions subject to wildly varying

interpretations. Far from “specifically prescribing] a course of action,” Berkovitz, 486 U.S. at 536, 108 S.Ct. at 1958, or creating “clear duties incumbent upon the governmental actors,” Kennewick Irrigation Dist., 880 F.2d at 1026, the Guidelines at best left all relevant details to the discretion of government employees. At worst, the policy did not by its terms require the Army to do anything as to people in D.B.S.’s position. Because the policy was unclear and insufficiently specific, the Army retained discretion whether to identify and warn contacts and persons at risk.

Thus, we hold that the Guidelines were not a “specific mandatory directive” requiring defendant to notify D.B.S. of the risk that he had received infected blood. Berkovitz, 486 U.S. at 544, 108 S.Ct. at 1963. Plaintiffs argue, however, that even in the absence of a duty to warn specifically created by regulations, § 2680(

a) does not apply because “notification ... [is a] matter[ ] of professional medical judgment.” First, we reiterate that labels do not control our analysis. Whether this was a matter of professional medical judgment does not alter the fact that our decision depends on applying the two-part test. Moreover, plaintiffs’ attempt to distinguish Bacon v. United States, 810 F.2d 827, 829 (8th Cir.1987), in which this court held that the EPA’s failure to warn contractors of dioxin contamination was protected by § 2680(a), is not persuasive.

In Layton, 984 F.2d at 1504, we construed Bacon and concluded that it found the government immune from suit because “where there is no regulatory requirement to issue a warning, the government may decide whether or not to warn persons as a matter of its own discretion.” Bacon and Layton are controlling here. Because there was no regulatory warning requirement on the defendant as to D.B.S., we cannot find that it lacked discretion whether to notify him. This is not to say that § 2680(

a) applies every time the government chooses not to warn when it has the discretion to make such a choice. For the government conduct must also satisfy the second part of the Supreme Court’s test; it must be grounded in public policy considerations. Again, the crucial issue is whether the defendant’s conduct is susceptible to policy analysis. The conduct at issue here is the government’s failure to issue a warning.

The decision of whether to identify and then warn thousands of military personnel about transfusions they underwent several years earlier, like the decision regarding screening procedures, implicates the competing concerns of safety and cost.

The government could have balanced the fact that identifying all those in D.B.S.’s position might have lowered the risks of transmission against the possibility that public awareness of AIDS and warnings from other sources might have obviated the need for notification from the Army, the fact that the risk of infection was statistically slight, the risk that military morale could be affected, and the judgment that scarce resources could be better allocated elsewhere.

See In re Consolidated United States Atmospheric Testing Litig., 820 F.2d 982, 997-98 (9th Cir.1987) (discussing the policy factors implicated in the government’s failure to warn participants in the United States atmospheric nuclear weapons testing program), cert. denied, 485 U.S. 905, 108 S.Ct. 1076, 99 L.Ed.2d 235 (1988). Moreover, as with the screening decision, this is true whether or not defendant in fact engaged in conscious policy-balancing.

The Layton court also cited the balancing of safety against cost as supporting its conclusion that the government’s decision in that case not to warn contractors of the dangers of felling trees on federal land was susceptible to policy analysis. See Layton, 984 F.2d at 1504. It distinguished several cases from other circuits, including Andrulonis v. United States, 924 F.2d 1210 (2d Cir.), vacated, — U.S. -, 112 S.Ct. 39, 116 L.Ed.2d 18 reinstated, 952 F.2d 652 (2d Cir.1991), cert. denied, — U.S. -, 112 S.Ct. 2992, 120 L.Ed.2d 869 (1992), in which courts held that § 2680(

a) did not protect a discretionary failure to warn. See Layton, 984 F.2d at 1504-05. The basis for the distinction was that in the other eases, “no consideration of competing values was implicated in the government’s decision not to issue a warning.” Id. at 1505. Thus, in Andrulonis, one scientist’s isolated decision not to warn another researcher about the known dangers of an experiment upon which they were both working did not implicate competing policy considerations. See Andrulonis, 924 F.2d at 1219.

The failure to issue a warning there did not rise to the level of a policy decision, but rather was merely “simple negligence.” Layton, 984 F.2d at 1505. In contrast, the defendant in Layton had real policy concerns to consider because “the more effort the Forest Service expended to discover dangers and to warn contractors of them, the greater the safety benefit but also the greater the cost to the government.” Layton, 984 F.2d at 1504.

Similarly, government employees here were faced with the decision of whether to devote significant resources to identify thousands of personnel across the country who fit certain criteria in order to warn them- about a risk that was statistically slight. This is a far cry from the isolated need for a single warning about a severe risk that could have been given in person to a party already identified in An-drulonis.

Plaintiffs argue that cost factors are always a consideration in government decisions and suggest that all government decisions would be immunized if cost alone “transform[ed] decisions into protected discretion.” The answer to this is that cost alone is not always sufficient to protect discretionary government conduct. The proper focus, rather, is on whether a decision is “of the nature and quality that Congress intended to shield from tort liability.” Varig Airlines, 467 U.S. at 813, 104 S.Ct. at 2764.

The issue of whether conduct is truly discretionary and whether there are real and competing policy considerations implicated is what separates Layton from Andrulonis. While we intimate no opinion as to precisely where this line should be drawn, suffice it to say that here defendant’s conduct was on the protected side of the line. In the end, what plaintiffs challenge in this case are the choices the government made. The defect in plaintiffs’ claims is that they require this court to engage in precisely the type of “judicial ‘second-guessing’ ” that the discretionary function exception was designed to avoid.

By so holding, we in no way endorse the decisions defendant made. We find only that they are the type of decisions that Congress intended to immunize from suit. III. CONCLUSION We hold that § 2680(

a) bars both of plaintiffs’ claims. We therefore affirm the district court’s order granting

summary judgment in favor of defendant. . Plaintiff C.R.S. died on June 26, 1993. . The Honorable Harry H. MacLaughlin, Senior United States District Judge for the District of Minnesota. . The FDA is the federal agency responsible for overseeing and regulating blood banks. . The AABB is a trade organization for the blood banking industry. . In Feres v. United States, 340 U.S. 135, 71 S.Ct. 153, 95 L.Ed. 152 (1950), the Supreme Court held that claims brought by members of the military for injuries related to military service are barred. .

In 1974, the Commissioner of the FDA and the Assistant Secretary of Defense (Health and Environment) entered into a Memorandum of Understanding under which the military agreed to abide by FDA regulations applicable to private blood banks. This agreement suggests that even assuming arguendo that the MBPO lacked authority to act on its own, it was indeed required to adopt FDA/AABB screening procedures. We need not decide this issue, however, in light of our conclusion that the MBPO had authority to make the screening decision. . Subsection B provided: B.

Identification of Contacts and Persons at Risk Contacts of HTLV-III-infected individuals and persons at risk of developing infection will be identified through the interview of HTLV-III-infected persons or through self-referral. These persons will be notified by local health authorities, usually Preventive Medicine Services or the installation POCs. They will be counseled concerning the reasons for the notification and advised to undergo testing to determine HTLV-III antibody status.

Active duty members will be required to have tests performed; dependents and civilians will be advised and encouraged to undergo testing. Appellants' App. at 201. The

section following subsection B is also relevant: C. Preventive Medicine Interview and Counseling 1. Interview Contacts and others that are identified as being at high or moderate risk of contracting HTLV-III infection will need to be interviewed.... Id. This provision apparently required that contacts and persons at risk identified pursuant to subsection B be interviewed. . This assumes that D.B.S. was not a “contact” of Donor 3903. Even if we classify D.B.S. as a contact, however, an interview of Donor 3903 •would not have yielded D.B.S.'s name.

Such an interview at most would have disclosed that Donor 3903 donated blood at MACH in August 1983. To identify D.B.S., the Army would then have had to trace Donor 3903's infected blood to D.B.S. The, general language of subsection B did not impose such a duty. See Berkovitz, 486 U.S. at 536, 108 S.Ct. at 1958. In any event, it is also unclear what “the interview of HTLV-III-infected persons” meant: it could have meant that the Army was supposed to interview people it already knew were infected or that the Army was supposed to test active duty members and then interview those who tested positive.

Alternatively, it could have meant that the Army was supposed to test anyone who had been in the Army since 1980 and then interview those who tested positive. Because this interview provision lacked clarity and specificity, it did not remove discretion from Army employees. See id. Thus, subsection B did not require the Army to interview Donor 3903 or any other "HTLV-III-infected persons.” . We note that D.B.S. was a member of the National Guard until 1987. . The risk was statistically slight at least as to people who received blood transfusions between 1980 and 1986.

The risk was of course higher for those in other groups within the Guidelines’ “High Risk Category,” such as intravenous drug users.