Patients at the Iowa Civil Commitment Unit for Sex Offenders (CCUSO) filed a complaint under 42 U.S.C. § 1988 challenging the placement of video cameras in CCUSO restrooms, and moved for a preliminary injunction to stop their use. The district court ^[1] denied the motion as to cameras in the “dormitory style restrooms” (restrooms with multiple toilets, showers and sinks) but granted a preliminary injunction ordering that cameras in the “traditional style bathrooms” (bathrooms with a single toilet, sink, and shower) be pointed at the ceiling or covered with a lens cap. The administrators of CCUSO appeal and we affirm.

The plaintiffs are civilly and involuntarily committed to CCUSO, a secure facility meant for the “control, care, and treatment of ... personfs] determined to be ... sexually violent predator[s].” See Iowa Code § 229A.7(7). After a reported sexual assault and an instance in which “a patient with a serious communicable disease engaged in consensual sex with another patient whom he did not inform about his condition,” the administrators installed cameras in all bathrooms, including those that are, as described by the district court, “similar to what people commonly have in their home with a toilet, shower, and sink in a smaller room.” Some of these single-user restrooms have doors while others have what the administrators describe as “T-junction[s]” in place of doors. Some with doors can be locked from the outside[*372] by staff but not by users inside the restroom, and two in a non-residential area can be locked by patients from inside; no rule prohibits multiple patients from being inside a single-user restroom at the same time. The administrators stated in the district court that the facility has at least some cameras that record “comings and goings to and from the restroom but not inside the restroom.”

While cameras in the common areas of CCUSO are monitored, those inside the single-user restrooms are not; instead, they record images that are generally erased within 14 to 21 days when the system records over them. The recorded images are “masked” so that most of the subjects’ bodies are covered with a black box, though several senior administrators have the ability to unmask the images if necessary for an investigation. The plaintiffs assert that the cameras placed in the single-user bathrooms are an unreasonable violation of their right to privacy, but the administrators contend that the cameras are needed to ensure the security of those committed to the institution and that the district court, when granting the injunction, failed to show sufficient “deference” to their “judgment” as “qualified professionally],” see Youngberg v. Romeo, 457 U.S. 307, 322-23, 102 S.Ct. 2452, 73 L.Ed.2d 28 (1982).

We review the district court’s grant of a preliminary injunction for an abuse of discretion and its factual findings for clear error. Barrett v. Clayeomb, 705 F.3d 315, 320 (8th Cir.2013). A district court should evaluate four considerations when determining whether to issue a preliminary injunction: the threat of irreparable harm to the movant, the balance between this harm to the movant and the harm an injunction will cause other parties, the probability that the movant will prevail on the merits, and the public interest. Roudachevski v. All-American Care Ctrs., Inc., 648 F.3d 701, 706 (8th Cir.2011).

Because the probability of “[success on the merits has been referred to as the most important of the four factors,” see id., we consider it first. The Fourth Amendment protects persons against unreasonable searches and seizures by the government. See, e.g., Brigham City v. Stuart, 547 U.S. 398, 403, 126 S.Ct. 1943, 164 L.Ed.2d 650 (2006). The administrators contend that no search or seizure occurs if nobody views the video images (though the record shows that they have viewed some of the videos in the past) and, if there is a search or seizure here, it plainly meets the Fourth Amendment reasonableness requirements.

A search occurs under the Fourth Amendment when, as relevant here, “the government violates a subjective expectation of privacy that society recognizes as reasonable.” Kyllo v. United States, 533 U.S. 27, 31-33, 121 S.Ct. 2038, 150 L.Ed.2d 94 (2001); see also United States v. Jones, — U.S. -, 132 S.Ct. 945, 950, 181 L.Ed.2d 911 (2012). “[I]nvoluntarily civilly committed persons retain the Fourth Amendment right to be free from unreasonable searches that is analogous to the right retained by pretrial detainees.” Beaulieu v. Ludeman, 690 F.3d 1017, 1028 (8th Cir.2012). Although the expectation of privacy shared by involuntarily civilly committed persons and pretrial detainees is of a “diminished scope,” see Bell v. Wolfish, 441 U.S. 520, 559, 99 S.Ct. 1861, 60 L.Ed.2d 447 (1979), neither our court nor the Supreme Court has ever outlined exactly what expectation of privacy these persons reasonably have, outside of our holding that detainees do not have a reasonable expectation of privacy in their jail cells, see United States v. Hogan, 539 F.3d 916, 923 (8th Cir.2008).[*373] We believe, though, that single-person bathrooms (intended for functions “ ‘traditionally shielded by great privacy’ ”) are inherently different from cells, and that a civilly committed person has a reasonable expectation of privacy in a single-person bathroom when there is no immediate indication that it is being used for purposes other than those ordinarily associated with bathroom facilities. See Vernonia School Dist. m v. Acton, 515 U.S. 646, 658, 115 S.Ct. 2386, 132 L.Ed.2d 564 (1995) (quoting Skinner v. Railway Labor Executives’ Assoc., 489 U.S. 602, 626, 109 S.Ct. 1402, 103 L.Ed.2d 639 (1989)); cf. United States v. Hill, 393 F.3d 839, 841 (8th Cir.2005). We therefore believe that by capturing images of patients while they occupy single-user bathrooms, CCUSO violated its patients’ reasonable expectation of privacy, thus conducting a search under the Fourth Amendment, irrespective of whether there is some chance that those images will not be viewed, see Kyllo, 533 U.S. at 31-33, 121 S.Ct. 2038.

After we determine that the government’s action is indeed a search, as we did here, we apply a so-called “balancing test” to determine whether the search or seizure of involuntarily committed individuals is reasonable, considering the “ ‘scope of the particular intrusion, the manner in which it is conducted, the justification for initiating it, and the place in which it is conducted.’ ” See Serna v. Goodno, 567 F.3d 944, 949 (8th Cir.2009), cert. denied, 558 U.S. 972, 130 S.Ct. 465, 175 L.Ed.2d 312 (2009) (quoting Bell, 441 U.S. at 559, 99 S.Ct. 1861). We “must defer to the judgment of correctional officials unless the record contains substantial evidence showing their policies are an unnecessary or unjustified response to problems of institutional security.” Beaulieu, 690 F.3d at 1029 (internal quotation marks and citation omitted).

The administrators point to our precedents allowing strip searches of civilly committed individuals for the purpose of finding contraband, see id. at 1027-30; Serna, 567 F.3d at 947-48, 952-56, and argue that their placement of the video cameras in the current circumstances is as necessary and justified as the strip searches in those cases. But the justification for those searches is quite different from the one offered for the video cameras here. In the cases the administrators rely on, the searches were meant to prevent dangerous contraband from entering the facilities, an immediate risk that was most easily prevented by strip searches. And we cautioned in Serna that if “the triggering evidence had been a prohibited but relatively benign object, ... it would have seemed, on balance, less reasonable to move quickly towards a method of searching that is so highly and personally invasive,” id. at 951. Here, however, neither the “triggering evidence” nor the means to prevent its harms are analogous to those in the earlier cases. Although unmonitored cameras may sometimes deter illicit behavior and help with investigations, they do not provide the administrators with immediate alerts concerning patient safety or directly prevent assaults or dangerous acts. They are instead an after-the-fact investigative tool: By the time the video is viewed, the harm has already happened.

While we do not “strictly apply a ‘less intrusive means’ test” for searches, see Beaulieu, 690 F.3d at 1029, we consider the availability of less intrusive techniques when assessing the reasonableness of a challenged procedure, see Serna, 567 F.3d. at 955. Here we are confident that there are less intrusive methods that the administrators can use to prevent various illicit activities by patients. As the district court suggested, and the administrators concede, allowing the doors of the bath[*374] rooms to be locked from the interior could significantly reduce the risk of assaults. The administrators contend that allowing patients to lock the doors would do “nothing to address sexual acting out,” reduce the ability of staff to monitor the bathrooms, and could delay medical assistance should a patient harm himself. We think that these assertions are unavailing. The staff can have keys to unlock the bathrooms, and we do not see how their having to unlock the door, rather than just opening it, will make monitoring the bathroom interiors significantly harder for the staff or create a delay in responding to medical emergencies. We also believe that the bathrooms that use “T-junctions” rather than doors, which are presumably easier to monitor (both visually and aurally), are also amenable to less-intrusive safety measures, such as a rule-change prohibiting multiple users from entering the restroom at the same time. Because the cameras in the “common areas” are monitored in real time, we think that a policy that would have staff investigate immediately when more than one person enters a single-user restroom would help prevent both assaults and consensual activities more efficiently than relying on the deterrence function of the cameras that are not monitored. We therefore see no error in the district court’s determination that the plaintiffs showed the requisite probability of success on their Fourth Amendment claim.

We conclude, moreover, that the district court did not clearly err, see Rogers Group, Inc. v. City of Fayetteville, Ark, 629 F.3d 784, 790 (8th Cir.2010), by finding that the plaintiffs established a threat of irreparable harm. “In order to demonstrate irreparable harm, a party must show that the harm is certain and great and of such imminence that there is a clear and present need for equitable relief.” Iowa Utilities Bd. v. F.C.C., 109 F.3d 418, 425 (8th Cir.1996). In their affidavits, the plaintiffs aver that the placement of the cameras cause them several kinds of irreparable harm: In addition to experiencing the general discomfort and embarrassment that might be expected from having a camera placed in the bathroom, the plaintiffs say that they are uniquely sensitive to the camera placement as several of them suffered significant sexual abuse during their lives and that having the cameras in these bathrooms interferes with their treatment and thus their potential release from civil commitment. After reviewing the affidavits, the district court found that the plaintiffs had “persuasively reported] significant on-going trauma and distress related to the placement of cameras in their bathrooms, which they, in sometimes eloquent and disturbing detail, equate with their previous experiences of victimization.”

The district court also determined that the patients’ interests in this case outweighed injuries that the injunction would inflict on other parties. The administrators contend that the injunction reduces patient safety in the CCUSO and assert that the district court did not give enough weight to their interest in preventing consensual sexual encounters among the patients. The administrators also suggest that there are no other reasonably effective safety measures than the video cameras. But we note again that while they might deter unsafe behaviors, the video cameras provide no immediate safety alert to the administrators and only provide evidence of past infractions. And, as we have already pointed out, we are confident that the administrators can find less-intrusive methods to protect the patients’ safety. While we recognize the important interests in upholding the rules of the CCUSO, including its prohibition on sexual activity,[*375] the district court did not abuse its discretion in its balancing.

We must also consider how an injunction would affect the public interest. There is certainly a “significant” public interest in keeping these types of institutions “safe and orderly.” See Serna, 567 F.3d at 954 (internal quotation marks and citation omitted). But we discern no error in the district court’s determination that in the circumstances of this case there is a greater public interest in protecting the Fourth Amendment rights of the patients and their “personal privacy and dignity against unwarranted intrusion by the State,” see Schmerber v. California, 884 U.S. 757, 767, 86 S.Ct. 1826, 16 L.Ed.2d 908 (1966).

For the reasons indicated, we conclude that the district court did not abuse its discretion in granting a preliminary injunction, and we affirm.