FLAUM, Circuit Judge.
Defendants James O’Grady, Sheriff of Cook County, and Spencer Leak,1 Executive Director of the Cook County Department of Corrections (“Department”) appeal the decision of the district court enjoining them from continuing mandatory urinalysis screening for all employees of the Department. For the reasons stated below, we find the injunction to be over-broad and remand the case for modification of the injunction consistent with this opinion.
[1191]*1191I. Facts and Proceedings Below
Defendant O’Grady is the Cook County, Illinois, Sheriff, the highest-ranking law enforcement officer of the County. The Illinois General Assembly has included within his office the Cook County Department of Corrections. Ill.Rev.Stat. ch. 125 ¶ 202. Among the powers delegated to the Sheriff is the authority to appoint the Department’s Executive Director, defendant Leak, who is responsible for the management and operation of the Department. The Executive Director’s authority extends to the employees and activities of this facility, including the formulation and implementation of personnel policies and regulations for the Cook County Jail. Pursuant to the authority of their offices, the individual defendants ratified and implemented Departmental General Order 9.8A, which provides for an annual mandatory urine testing program (“AMUT”) effective August 26, 1986, for all employees of the Department.2
A. The Urinalysis Test
The AMUT program, as its full name suggests, is a systematic urinalysis program designed to detect the use of illegal drugs by Department correctional employees. Under the program every correctional officer and correctional supervisor3 of the Department would be compelled to produce a urine specimen once each year for chemical analysis. A correctional officer would have no advance notice of the specific day on which she would be required to produce a urine specimen. Officers scheduled for testing would be advised during the roll call at the beginning of their shift that they must produce a urine specimen that day. They would then be taken to a holding area where they would remain until they could produce a urine specimen. Under the program, they would have until the close of the shift to produce a specimen. If not, the unsuccessful, presumably unwilling, officer could be terminated.
Immediately prior to producing a specimen, officers would be requested, but not required, to specify medical information relating to their drug and medical history. The purpose of the inquiry is informational and precautionary; in the event of a false positive urinalysis result, the officer would have on record prior recorded responses to help explain that result.
Each officer to be tested would be accompanied to a restroom by a testing administrator of the same sex. Prior to her admission, the restroom would be searched for any objects, articles, or materials which may corrupt the specimen. The officer would then be subjected to a pat-down search. Thereafter, the administrator would hand her a cup to urinate into and admit the officer to the restroom for that purpose. The testing administrator would maintain “some visual contact” of the officer from a “discreet” distance. The general order does not require or recommend direct observation of the flow of urine from the officer’s body into the cup unless the investigator has reason to believe that the officer might attempt to falsify her sample by, for example, switching samples or substituting water for the urine specimen. Ordinarily, women officers would be allowed to produce a specimen within a stall with the door closed, while male officers would use a urinal.
The officer’s specimen would be placed in two separate containers, sealed, and sent to the Department. Then one container would be opened and the specimen analyzed for the presence of marijuana, cocaine, and opiates through the Enzyme Multiplied Immunoassay Test (“EMIT”). [1192]*1192If the EMIT test is positive, the opened container would be resealed and, along with the unopened container, sent to the Met-Path Laboratories in Wood Dale, Illinois, for a second analysis utilizing the Gas Chromatography-Mass Spectroscopy test (“GCMS”).4 If the second test also proved positive, the officer would have the option of having another GCMS test performed upon the sample.
If the second GCMS failed to confirm the previous positive results, no action would be taken against the officer. If, however, the second GCMS test produced the same positive result, or if the officer declined to have a second GCMS performed, she would be presented with the option of entering a treatment program for drug abusers or, if she declined to enter the treatment program, the Sheriff could file a complaint with the Cook County Police and Corrections Merit Board seeking termination of the officer. If the officer entered the drug treatment program she would be compelled to produce urine specimens for analysis twice each month for the next six months. The Sheriff could seek the officer’s discharge if any of these specimens produced a positive result, if the officer “fails to successfully complete the program,” or if following the completion of the drug treatment program any of the officer’s regular AMUT specimens produced a positive result.
B. The District Court’s Findings
Following the announcement of the planned implementation of the AMUT program, the plaintiffs initiated this lawsuit. They comprise a certified class of all correctional officers and correctional supervisors,5 approximately 1700 persons in total, who seek to have the Department enjoined from continuing the AMUT program. Premising their lawsuit on 42 U.S.C. §§ 1983, 1988, and 28 U.S.C. § 2201, they argue that the urinalysis testing violates the fourth amendment. They also seek a declaratory judgment that the program is unconstitutional.
Following a bench trial, Judge Getzen-danner issued a declaratory judgment that the AMUT program violates the fourth amendment and therefore enjoined the Department from implementing the AMUT program. Taylor v. O’Grady, 669 F.Supp. 1422 (N.D.Ill.1987). At trial and continuing here on appeal, the Department has advanced three interests to be served by the AMUT program. First, the Department argues that the AMUT program will maintain an unimpaired work force, a work force otherwise said to be infected to some degree with drug abuse. Second, the Department believes the program will prevent correctional officers from smuggling illegal drugs into the Cook County Jail and selling the drugs to the inmate population. Finally, the AMUT program is said to foster the public’s perception of the integrity of the correctional officers as law-abiding law enforcement personnel.
Based on exhibits and testimony relating to the incidence of both drug abuse and drug-trafficking by correctional officers, the district judge concluded that drug abuse or drug trafficking among the correctional officers occurred in rates similar to that of the general population. The court found support for Dr. Schnoll’s expert testimony concluding that alcohol, which the AMUT program would not de[1193]*1193tect, is the most common substance abused, and that the numbers of non-alcohol substance abusers in society at large and therefore at the jail were relatively low; Dr. Schnoll’s statistical determinations based on rates of abuse in the general population indicated that only three of 1700 correctional officers were chronic drug abusers and might possibly sell drugs to prisoners. The district judge also found that a serious drug smuggling problem was present within the prison, but she also accepted evidence that most of the smuggled drugs probably came from sources other than the correctional officers, such as the prisoners’ visitors.
As for the Department’s interest in ensuring that its work force is not impaired, Judge Getzendanner agreed that the planned testing procedure was indeed accurate, but concluded that a positive result merely indicated that an officer had abused a drug at some time, not necessarily while she was working. Moreover, based primarily on the testimony of Dr. Schnoll, Judge Getzendanner found that trained supervision combined with concomitant effective disciplinary procedures, was a less intrusive measure than the AMUT program that would effectively detect chronic drug abusers and those officers impaired while on duty, and consequently would sufficiently reduce the drug abuse problem at the prison.
With these findings of fact established, the district court then engaged in a fourth amendment analysis. While recognizing that the AMUT program may be justified in its inception as reasonably related to some legitimate government interests, the court determined that the program nevertheless failed to advance those interests significantly. The district judge reasoned that the AMUT program was unnecessarily intrusive since trained supervision, a far less intrusive method of promoting the Department’s interests, was at least as successful as urinalysis, and should detect the chronic drug abusers among the officer population. This would, the court said, maintain an unimpaired work force whose integrity should then become apparent to the public. According to the judge, the evidence does not show that a positive AMUT test would even be likely to reveal impairment on the job.
The court conceded that urinalysis might be reasonable to test those drug abusers picked up through trained supervision and also might be appropriate if confined to the Department’s law enforcement officers. Nevertheless the court concluded that the AMUT program as designed was too sweeping to pass the test of reasonableness under the fourth amendment. In addition to a declaratory judgment to that effect, the court enjoined the AMUT program “in the absence of a reasonable suspicion of the use of illegal drugs or the illegal use of drugs.” 669 F.Supp. at 1443.
The Department appeals the district court’s decision6 on a variety of grounds. It first argues that the decision is wrong on the merits and urges this Court to reverse that decision outright. Alternatively, the Department urges a remand based on erroneous findings of fact, abuses of discretion in excluding witnesses, and improper consideration of hearsay documentary evidence for its truth. Finally, the Department contends that a comment made in chambers by Judge Getzendanner and another made at trial indicated that she harbored a personal bias against urinalysis which resulted in an unfair trial.
II. Analysis
A. The Intervening Supreme Court Cases
The Supreme Court had not addressed the issue of urinalysis until after the oral arguments in this case. Following oral arguments and while this appeal was still pending, it issued decisions in companion cases, National Treasury Employees Union v. Von Raab, — U.S. -, 109 S.Ct. 1384, 103 L.Ed.2d 685 (1989), and Skinner v. Railway Labor Executives’ Ass’n, — [1194]*1194U.S. -, 109 S.Ct. 1402, 103 L.Ed.2d 639 (1989), which bear on our disposition of this case. In Von Raab, the Court determined that the United States Customs Service could compel urinalysis for those employees seeking transfers or promotions to certain specified positions within the Service. In Skinner, the Court similarly decided that the Federal Railroad Administration could require railroads to administer blood and urine tests to train workers involved in major railroad accidents, and could authorize railroads to administer breath and urine tests to workers who violate certain safety rules. Because of the significance of these decisions, we invited the submission of supplementary briefs from both sides, which were so provided.
While the import of both decisions is obvious, and general principles on the constitutional framework with which to analyze urinalysis programs such as the one before us can now be articulated with confidence, nonetheless factual distinctions exist from program to program. Those distinctions comprise the critical border between an employer’s reasonable intrusion upon its employees’ privacy under the fourth amendment and an unconstitutional foray. Our analysis begins with an overlay of the general principles and then examines the specific scenario of this case to determine upon which side of the border the AMUT program rests. Since defendants are Cook County and two of its officers, our discussion is limited to governmental testing.
A program which compels governmental employees to submit to urinalysis is a search within the meaning of the fourth amendment, for such tests invade reasonable expectations of privacy. Von Raab, 109 S.Ct. at 1390; Skinner, 109 S.Ct. at 1412-1413; see also Schaill v. Tippecanoe County School Corp., 864 F.2d 1309, 1312 (7th Cir.1988); cf. Schmerber v. California, 384 U.S. 757, 86 S.Ct. 1826, 16 L.Ed.2d 908 (1966). Consequently, such a drug testing program must meet the reasonableness requirement of the fourth amendment. Von Raab, 109 S.Ct. at 1390.
Although urinalysis programs are within the constrictions of the fourth amendment, neither a warrant, probable cause, nor individualized suspicion necessarily need be shown to justify them. Von Raab, 109 S.Ct. at 1390; Skinner, 109 S.Ct. 1413, 1414. Nor in fact does the government need to demonstrate that any suspicion at all, individualized or otherwise, exists; the government is not required to show that the group to be tested as a whole suffers from any incidence of drug abuse. Von Raab, 109 S.Ct. at 1391-1393, 1395 (“The mere circumstance that all but a few of the employees tested are entirely innocent of wrongdoing [no more than 5 out of the 3600 employees tested positive for drugs] does not impugn the program’s validity”).7 Instead, “where a fourth amendment intrusion serves special governmental needs, beyond the normal need for law enforcement, it is necessary to balance the individual’s privacy expectations against the Government’s interests to determine whether it is impractical to require a warrant or some level of individualized suspicion in the particular context.” Von Raab, 109 S.Ct. at 1390, citing Skinner, 109 S.Ct. at 1413-1414; see also Von Raab 109 S.Ct. at 1395 (“Where, as here, the possible harm against which the Government seeks to guard is substantial, the need to prevent its occurrence furnishes an ample justification for reasonable searches calculated to advance the Government’s goal.”); O’Connor v. Ortega, 480 U.S. 709, 107 S.Ct. 1492, 1499, 94 L.Ed.2d 714 (1987); United States v. Place, 462 U.S. 696, 103 S.Ct. 2637, 2642, 77 L.Ed.2d 110 (1983).
[1195]*1195In both Von Raab and Skinner the Supreme Court recognized some special governmental interests which warranted suspi-cionless drug testing. Perhaps of foremost importance to the Court was the governmental interest in preserving the safety of the public as well as fellow employees from the threat of an impaired employee with access to dangerous instrumentalities. Customs agents who are involved in drug interdiction, carry firearms, and train workers “discharge duties fraught with such risks of injury to others that even a momentary lapse of attention can have disastrous consequences.” Skinner, 109 S.Ct. at 1419; Von Raab, 109 S.Ct. at 1393. The Court majority also noted that the government’s “compelling interest in protecting truly sensitive information” could in some circumstances be a basis for suspicionless drug testing of employees with access to such information. Von Raab, 109 S.Ct. at 1396. Finally, the government’s interest in ensuring that “front-line interdiction personnel are physically fit, and have unimpeachable integrity and judgment” was recognized in Von Raab as compelling.8 109 S.Ct. at 1393.
These cases also noted that the availability of less intrusive alternatives to urinalysis does not impact upon the constitutionality of the program being challenged. Skinner made clear that judges are not to engage in a post-hoc exercise of imagining some other path of conduct the government could have taken. Instead, courts should analyze the challenged program in and of itself to determine whether it comports with the fourth amendment. “[T]he reasonableness of any particular government activity does not necessarily or invariably turn on the existence of alternative ‘less intrusive’ means.” Skinner, 109 S.Ct. at 1419 n. 9, (quoting Illinois v. Lafayette, 462 U.S. 640, 647, 103 S.Ct. 2605, 2610, 77 L.Ed.2d 65 (1983)).
As a consequence of Skinner and Von Raab, the issue of governmental suspicion-less urinalysis for drug screening essentially rests on a balancing test: the interests of the government balanced against the privacy interests of the employees. And the privacy interests of the employees are not simply composed of the desire to be free from mandatory urinalysis, but also concern the intrusiveness of the particular program in issue. Thus while urinalysis may be within the government’s prerogative in a given circumstance, the manner in which the program is carried out may be so unnecessarily intrusive as to render it constitutionally intolerable.
B. Preliminary Evidentiary and Factual Challenges
Before considering the constitutionality of the AMUT program under the fourth amendment, several procedural claims rendered irrelevant by Von Raab and Skinner should be dismissed. Three of the Department’s challenges on appeal relate either to evidence suggesting that the Department is suffering from a drug abuse problem among its employees, or to Judge Getzen-danner’s finding that no serious drug abuse problem in fact existed. Specifically, the Department complains that (1) Judge Getzendanner, as a sanction, abused her discretion by precluding the Department from calling eighteen lay witnesses with personal knowledge of drug abuse at the Department of Corrections and an expert witness who held opinions on this matter; (2) she considered hearsay evidence relating to the number of reported incidents involving drugs for its truth in contravention of a previous agreement that the evidence be admitted for a limited purpose; and (3) her ultimate finding that there was no serious drug abuse problem at the Department was clearly erroneous.
[1196]*1196All of these matters regard the degree of suspicion or lack thereof which would support the government’s decision to engage in a urinalysis program. But as already pointed out, both Von Raab and Skinner supported urinalysis programs without any level of individualized suspicion nor even suspicion relating to the group as a whole to be tested. It was explicitly conceded in Von Raab that the Customs Service did not have a drug problem. Consequently, whether there exists a drug problem at the Department is not pertinent; that component has been eliminated in favor of a balancing test. Accordingly, the three challenges are irrelevant.
C. The Constitutionality of the AMUT Program
In both Von Raab and Skinner, the Court chose not to articulate a concrete analytical framework for this type of case. Consequently, the test of constitutionality still depends on a balancing of the individual’s privacy expectations against the government’s interest to be served by the test. Courts must therefore scrutinize the procedures used by the government and the specific interests allegedly served by testing a particular category of employees, as the D.C.Circuit has done on two recent occasions. American Federation of Government Employees v. Skinner, 885 F.2d 884 (D.C.Cir.1989); National Federation of Federal Employees v. Cheney, 884 F.2d 603 (D.C.Cir.1989). Accordingly, we carefully scrutinize the facts of this case before balancing the interests.
1. The Department’s Interests
The Department advances three interests in randomly testing all of its employees: fostering the public’s perception of the integrity of its work force; maintaining an unimpaired, physically fit work force; and preventing the smuggling of drugs to prisoners by correctional officers. Our task is to weigh these interests based on the record before us.
The Department clearly has some interest, as all governmental employers do, in ensuring a law-abiding and drug-free work force. Indeed, public employers may have a higher interest in the integrity of their work force than the private sector as they legitimately may expect their employees to be the standard bearers of society. Drug testing may promote the public’s perception of the integrity of its work force by making a public statement that no drug-users are employed at the Department.
A generalized interest in the integrity of the work force, however, is not enough to overcome the privacy interests at issue. Such a result would be at odds with Von Raab. The Von Raab Court noted that “[ujnlike most private citizens or government employees in general, employees involved in drug interdiction reasonably should expect effective inquiry into their fitness and probity.” 109 S.Ct. at 1394 (emphasis added). See also Harmon v. Thornburgh, 878 F.2d 484, 490 (D.C.Cir.1989) (“Von Raab it seems to us, suggests that federal employment alone is not a sufficient predicate for mandatory urinalysis.”). Moreover, the Court remanded that portion of the case dealing with the testing of employees with access to classified information because the class of employees tested was drawn too broadly. 109 S.Ct. at 1397. This remand would be superfluous if all government employees could be subject to mandatory testing based on the integrity rational. The governmental interest in the integrity of its work force, therefore, is not sufficiently strong to overcome the privacy interests of their employees.
The Department’s remaining interests are more substantial, at least as regards certain employees. Foremost is the department’s security interest in maintaining the jail. Cook County Jail houses criminals charged with a variety of crimes, from misdemeanors to murder and many of the prisoners may be violent. As one of the largest and most populous county jail facilities in the country, it is “fraught with serious security problems.” Bell v. Wolfish, 441 U.S. 520, 559, 99 S.Ct. 1861, 1884, 60 L.Ed.2d 447 (1979). Drug impairment by a guard, who may very well be armed, carries rather extreme and immediate dangers. See McDonell v. Hunter, 809 F.2d [1197]*11971302, 1308 (8th Cir.1987) (The use of drugs by employees who come into contact with the inmates “poses a real threat to the security of the prison.”). The guards “discharge duties fraught with such risks of injury to others that even a momentary lapse of attention can have disastrous consequences.” Von Raab, 109 S.Ct. at 1393 (quoting Skinner, 109 S.Ct. at 1419.) Certainly, those employees who come into contact with potentially violent prisoners must remain as alert as possible and free from lapses due to drugs.
The governmental interest in preventing the smuggling of drugs to prisoners is similarly weighty. The presence of drugs in prison can lead to violence and the prospect of correctional officers selling illegal drugs to prisoners is an obvious security risk. A drug-induced episode carries with it the potential onslaught of violence by the impaired prisoner or escapee. Other prisoners, visitors, correctional officers, and the public are all put at risk by the presence of drugs in a jail. Not only are prisoners under the influence of drugs a threat to safety, but prison employees who sell to prisoners are subject to blackmail. While urinalysis will not directly lead to drug dealers, the district court expressly found that those officers who smuggle drugs into the prison are likely to be chronic drug abusers. Taylor v. O’Grady, 669 F.Supp. 1422, 1429, 1437 (N.D.Ill.1987). See also, McDonell, 809 F.2d at 1308. Mandatory testing of correctional officers, therefore, “serves special governmental needs, beyond the normal need of law enforcement.” Von Raab, 109 S.Ct. at 1390.
But the Department’s interests in avoiding the dangers resulting from an impaired work force or from drug smuggling are not furthered by testing all corrections officers. Its AMUT program compels urinalysis for all correctional officers from the Executive Director to the lowest jail guard. Yet egalitarianism has nothing to do with the fourth amendment analysis in issue. The Court in Von Raab similarly questioned whether the “Service [had] defined this category of employees more broadly than necessary to meet the purposes of the Commissioner’s directive.” 109 S.Ct. at 1397. The Court required that the category of employees to be tested “directly encompass” those employees implicated by the governmental interests.
Applying this criteria, it is apparent that only those employees in contact with prisoners are implicated by either of the two stronger governmental interests here. Arguably the Department has less interest in the constant alertness and preparedness of these “non-contact” employees. Nor is there a threat of smuggling by employees not in contact with prisoners. Indeed, employees not in contact with prisoners are indistinguishable from other nonprison governmental employees who, as discussed above, are not subject to suspicionless, warrantless searches. Since those officers with only administrative or clerical duties or otherwise lacking contact with the jail population do not threaten claimed dangers if impaired while on duty, and since the record does not show they are able to smuggle drugs to the prisoners, the Department gains nothing by testing them.
2. The Privacy Interests of the Correctional Officers
The next task is to measure the individuals’ privacy rights based on the nature of the intrusion and their expectation of privacy. This court, along with others, has previously noted the nearly universally held expectations of privacy attached to the act of urination. See Schaill, 864 F.2d at 1312; Lovvorn v. City of Chattanooga, 846 F.2d 1539, 1543 (6th Cir.1988) (“There are few other times where individuals insist as strongly and universally that they be let alone to act in private.”); Von Raab, 816 F.2d 170, 175-176 (5th Cir.1987) (“There are few activities in our society more personal or private than the passing of urine.”), affirmed, — U.S. -, 109 S.Ct. 1384, 103 L.Ed.2d 685 (1989); Capua v. City of Plainfield, 643 F.Supp. 1507, 1513 (D.N.J.1986) (“Urine ... is normally discharged and disposed of under circumstances that merit protection from arbitrary interference.”). All urinalysis programs implicate serious privacy concerns regardless of how carefully tailored the program is designed. [1198]*1198Being compelled to urinate while under monitor is intrusive and often embarrassing and uncomfortable.
The intrusion into the expectations of privacy by the AMUT program, however, is not significantly different in magnitude from the intrusions approved by the Supreme Court in Von Raab or in Skinner. The AMUT program is a systematic and comprehensive program, testing all employees once, and only once, a year.9 The greatest difference between the AMUT program and the testing approved in Von Raab and Skinner, is that in those programs testing occurred only on the happening of a particular event. The eventual purpose of the testing in Von Raab, however, was to have tested all employees in certain, sensitive positions, as is the case in the AMUT program. Furthermore, in Skinner, testing occurred on the happening of a random event (accidents of a certain type), and an employee could be subject to testing on any given day without warning.
The plaintiffs confuse the fact that employees are not told until the day of their testing with randomness, arguing that Von Raab does not support random testing. The program, however, is not random; it is systematic and comprehensive, testing each officer once each year. Even admitting that the testing is perceived as random by the officers, we find no significant difference between this case and Von Raab or Skinner. Random testing has been upheld in at least two other circuits. McDonell v. Hunter, 809 F.2d 1302 (8th Cir.1989); American Federation of Government Employees v. Skinner, 885 F.2d 884 (D.C.Cir.1989). As the D.C. Circuit stated in American Federation, “[wjhile it is true that random testing may increase employee anxiety and the invasion of the subjective expectations of privacy, it also limits discretion in the selection process and presumably enhances drug-use deterrence.” Id.
The plaintiffs would also have us find fault in the AMUT program because it offers no alternative to testing other than termination. Yet in Von Raab, the drug testing was mandatory if the individual wished to receive the promotion or to be hired. Certainly, no painless alternative was available and none was necessary. And in Skinner, employees who refused testing, while entitled to a hearing, were not allowed to occupy their previous jobs for nine months. Moreover, the AMUT program only states that refusal to produce a urine sample is grounds for termination. Actual termination remains discretionary, with power to make the decision vested in the Sheriffs Office which supervises the Department.
The protection of individual privacy during the testing is also similar to that found in Von Raab. Under the AMUT program, the guard to be tested may produce the urine sample in a stall with the door closed.10 Actual physical observation of the flow of urine from the body into the cup is not required or recommended. The investigator running the test may not require the officer to produce the specimen in open view unless she has reason to believe that the person giving the sample is going to attempt to impugn the veracity of the test. This is similar to the testing in Skinner, where the observation of the flow of urine was also not required though not prohibited. Indeed, where there is reason to suspect tampering with the sample, [1199]*1199there may be individualized suspicion, justifying the necessarily greater intrusion.
The fact that the test is not performed by a medical officer is not sufficient to warrant a different finding from Von Raab.11 So long as the procedures used are the same, the collection of samples by non-medical personnel does not significantly differ from collection by trained personnel. In fact, the testing approved in Von Raab provided that the sample be collected by non-medical personnel in public restrooms. See also American Federation of Government Employees v. Skinner, 885 F.2d 884 (D.C.Cir.1989) (testing by non-medical personnel upheld.).
Finally, the safeguards in the AMUT program are designed to minimize any possible error and stigma for officers who test positive. Personnel have the opportunity to specify medical information relating to their substance use prior to testing, which prevents charges of recent fabrication. In addition, the most accurate of the urinalysis tests available, the GCMS test, is used. See Note, Employee Drug Testing, 74 VA. L.REV. at 987-989 (1987). Moreover, if an employee tests positive, she is not subject to termination. Instead, she must enroll in a drug treatment program, reducing the stigma of a positive test.
In sum, these tests are not significantly different from those in Von Raab, and Skinner. While all urinalysis is intrusive, these tests are not more intrusive of the individual’s privacy rights than those already approved by the Supreme Court.
3. Balancing the Interests
The remaining task is to balance the Department’s interest in the AMUT program with the intrusion into the individual’s privacy interests. The Supreme Court required that the category of employees directly encompass those who implicate the governmental interest. As noted above, only those employees who come into regular contact with prisoners, or who have opportunities to smuggle drugs to prisoners are implicated by the Department’s interests.
As to the other, administrative personnel, the governmental interest is not sufficient to overcome their privacy interests. The only legitimate interest advanced by the Department for testing these employees is to protect the general integrity of the work force. This is not enough. A generalized interest in the integrity of the work force will not sustain suspicionless urinalysis testing, and therefore the injunction is sustained as to these officers.
As for those officers who come into regular contact with prisoners, however, the injunction cannot be sustained in the face of Von Raab. As noted above, the Department has two substantial interests in testing these officers. Prisoners can be potentially violent and many of the guards are armed. A momentary lapse of alertness could lead to irreparable harm. In addition, the intrusion into the officer’s privacy is designed to be minimal and is in line with the intrusion authorized by Von Raab. Balancing these interests, we are compelled by Von Raab to find the AMUT program reasonable as to these officers.
This decision is consistent with the balances struck by other courts. In Harmon v. Thornburgh the D.C. Circuit considered a urinalysis program designed to test on a random basis Department of Justice employees in certain sensitive positions. 878 F.2d 484 (D.C.Cir.1989). The court held that the government’s interests in the integrity of the work force and public safety were not sufficient to allow the testing, but that the government could test those employees who had access to sensitive information. While public safety was not held [1200]*1200to be a sufficient interest in that case, the risks at issue were considerably different from the risks in this case. Indeed, the Harmon court itself distinguished that case from one such as ours:
The sort of indirect risk [at issue here], however, is wholly different from the risk posed by a worker who carries a gun or operates a train_ The public safety rationale adopted in Von Raab and Skinner focused on the immediacy of the threat. The point was that a single slip-up by a gun-carrying agent or a train engineer may have irremediable consequences; the employee himself will have no chance to recognize and rectify his mistake, nor will other government personnel have an opportunity to intervene before the harm occurs.
878 F.2d at 491. In this case, as opposed to Harmon, there are immediate safety risks implicated by drug-impaired employees, and therefore, our holding is consistent with Harmon.
Our holding is also consistent with National Federation of Federal Employees v. Cheney, 884 F.2d 603 (D.C.Cir.1989). Cheney concerned the random testing of certain Army civilian employees who held “critical” jobs. The court, among other things, upheld the testing of civilian police and guards. Yet the interest in testing security guards in that case arguably was less than the Department’s interest in testing security guards in a prison setting. In addition, the testing upheld was random; a fact not present here. See also American Federation of Government Employees v. Skinner, 885 F.2d 884 (D.C.Cir.1989) (upholding the random testing by the Department of Transportation of a certain class of employees, largely made up of air traffic controllers, designated because of the “safety and security criticalness” of their jobs); Transport Workers’ Union of Philadelphia v. Southeastern Pennsylvania Transportation Authority, 884 F.2d 709 (3rd Cir.1989) (upholding the random urinalysis of mass transit workers); Seelig v. Koehler, N.Y.Sup.Ct.App.Div. 1st Dept., App.Div., 546 N.Y.S.2d 828 (1989).
The proper course for the Court to take in this instance is a remand to the district court for further fact-finding before modifying the injunction. It is impossible based on the record now before us to define with particularity all of the personnel in the jail who are exposed to inmates on a regular basis. In addition, it is not proper to uphold the entire injunction merely because a portion of the program is over-broad. As is evident from the analysis, large portions of the testing program are valid. By upholding the entire injunction, we would be delaying an admittedly valid program for years. Indeed, the Supreme Court, in Von Raab remanded a portion of the testing program for further fact-finding by the district court rather than invalidating the entire program. This Court has also remanded cases in such circumstances. See e.g., Dreher v. Sielaff 636 F.2d 1141, 1145 (7th Cir.1980); Zbaraz v. Quern, 596 F.2d 196 (7th Cir.1979). While it is often appropriate to let the agency reformulate its plan, here, such a course would delay imposition of testing for years. In addition, the Department will lose no freedom to reformulate its plan under the modified injunction as opposed an invalidated program: either way it must comply with constitutional limits and the injunction merely sets those limits.
D. Recusal of District Judge Getzen-danner
Finally, the Department contends that two remarks made by Judge Getzendanner indicate that she harbored a personal bias against urinalysis generally and the Department as a party, so that she should have recused herself. Section 455(a) of the Judicial Code, 28 U.S.C. § 455(a), requires a judge to recuse herself in any proceeding in which her impartiality might reasonably be questioned. Section 455(b)(1) requires disqualification when the judge has a personal bias or prejudice concerning a party or personal knowledge of disputed evidentiary facts. Recusal under Section 455 is self-executing; a party need not file affidavits in support of recusal and the judge is obligated to recuse herself sua sponte under the stated circumstances. United [1201]*1201States v. Balistrieri, 779 F.2d 1191, 1202 (7th Cir.1985). Appellate review of such claims is de novo, and the standard of proof is whether a reasonable person would be convinced that the judge was biased. Id. The Department has not made the necessary showing that this trial judge was in any way biased or prejudiced.
The Department first made this contention in its posttrial motions to successor Judge Leinenweber. Judge Getzendanner had no opportunity to address this challenge. Nevertheless, the Department argues that under Balistrieri, Judge Getzendanner should have recused herself sua sponte. At the start, Section 455(a) is unavailable to the Department because “mandamus is the only method of correcting a judge’s refusal to recuse himself under 28 U.S.C. § 455(a).” Union Carbide Corp. ¶. U.S. Cutting Service, Inc., 782 F.2d 710, 713 (7th Cir.1986); see also Durban v. Neapolitan, 875 F.2d 91 (7th Cir.1989).
As to Section 455(b), the genesis of this belated claim is in two remarks made by Judge Getzendanner which the Department contends indicated her bias or prejudice. First, at a pre-trial conference in chambers, the judge allegedly stated, “Isn’t there any other way we can do it? I hate urine testing. I had to give urine samples when I was pregnant, and I hated it.”12 Second, during the trial when Sergeant MeNeal was describing her experience with the urinalysis program which preceded AMUT, the judge commented, “We are going through this [testimony] because there are no women judges on the Court of Appeals. But I would certainly know how she gave a urine sample....”
Neither of these remarks presents compelling evidence that the district judge harbored a personal bias against urinalysis or against a particular party. That Judge Getzendanner dislikes urine testing is certainly not indicative of a disqualifying bias; undoubtedly few people enjoy the process envisioned by the Department. Moreover, the judge was apparently exploring whether settlement was possible by inquiring as to the existence of an alternative method. As for her comment regarding the absence of women judges on this Court, no bias or lack of impartiality can reasonably be drawn from that. The remark was likely intended to explain the necessity of developing the record for the benefit of the male judges on this Court, despite female witness McNeal’s discomfort in testifying about such intimate matters. Moreover, it should be noted that Judge Getzendanner did approve an earlier urinalysis program for police officers in Harris v. Washington, No. 84 C 8812, unpublished op., 1985 WL 17506 (N.D.Ill. Feb. 6, 1985), which she carefully distinguished in her opinion in this case 669 F.Supp. at 1440-1444. The overall record clearly demonstrates that Judge Getzendanner exhibited no disqualifying bias or prejudice. The challenged remarks from the record are too inconsequential to mandate disqualification.
III. Conclusion
We vacate the present over-broad injunction and remand the cause to Judge Leinen-weber to modify the relief consistent with this opinion. The defendants should be enjoined from testing those employees who have (1) no regular access to inmate population, (2) no reasonable opportunity to smuggle drugs into the inmate population, and (3) no access to firearms.
Unless the parties can promptly agree on a more limited urinalysis program, the district judge is empowered to hear additional evidence confined to the scope of a substitute injunction. Through this procedure, the Department will be permitted to formulate a more reasonable testing method limited to an appropriate group of employees. Due to its importance, the district judge should proceed with the matter as expeditiously as possible. Circuit Rule 36 shall not apply.
[1202]*1202Judgment vacated and cause remanded for the purposes of fashioning a fresh decree in accordance with this opinion, our mandate to issue forthwith.