This study was performed at 2 tertiary acute care hospitals, Duke University Medical Center (753 beds) and the University of North Carolina Health Care (804 beds), from July 2011 through September 2012. The study protocol was reviewed by institutional review board committees at both institutions and determined to be exempt.

We performed an interventional study on a convenience sample of hospital rooms. Three pathogenic organisms were targeted: VRE, C. difficile, and Acinetobacter spp. Hospital rooms were identified using microbiological and infection control databases to search for patients placed on contact precautions as a result of colonization or infection with a target organism.

After a targeted room was identified, environmental cultures were obtained after patient discharge but before standard terminal room disinfection by environmental services personnel. A minimum of 5 environmental sites were cultured in triplicate from each room, using Rodac plates. The 5 environmental sites targeted for culture included the bedside rail, bedside table, chair arm, overbed table, and sink counter. When one of these surfaces was not available, supply carts were cultured. The toilet, shower floor, and floor adjacent to the toilet were also cultured in targeted rooms from which a patient with C. difficile infection was just discharged. Each environmental culture site was assessed and recorded as in either direct or indirect line of sight of the automated UV-C-light-emitting device. The automated UV-C-emitting device was then used in the room. Environmental cultures were repeated in triplicate from the same environmental sites, following application of UV-C light. Following these cultures, environmental services performed a standard terminal room disinfection per standard hospital protocol, and the room was made available for the next patient.

We sampled 229 environmental surfaces from the rooms of 39 patients during the 15-month study period. In total, 142 samples were obtained from 27 rooms of patients who were colonized or infected with VRE, 77 samples were obtained from 10 rooms of patients with C. difficile infection, and 10 samples were obtained from 2 rooms of patients with infections due to Acinetobacter. The median time for the UV-C vegetative cycle to be completed was 25 minutes (IQR, 20–35); the median time for the UV-C spore cycle to be completed was 45 minutes (IQR, 42–61).

The total number of CFUs of any type of pathogen detected on culture plates from all sampled environmental sites decreased from 28,642 to 2,444 following use of the UV-C device (1.07 log₁₀ reduction; Table 1). The median number of CFUs per sample decreased from 110 (IQR, 49–251) to 4 (IQR, 1–11) following use of the UV-C device (P < .0001). Similarly, the total number of CFUs of target organisms from all cultured environmental sites decreased from 1,488 to 66 following use of the UV-C device (1.35 log₁₀ reduction; P < .0001). A greater than 1 log₁₀ overall reduction was achieved for all 3 target organisms following use of the UV-C device. VRE CFUs decreased from 712 to 15 following use of the UV-C device (1.68 log₁₀ reduction; P < .0001), C. difficile CFUs decreased from 724 to 51 (1.16 log₁₀ reduction; P < .0001), and Acinetobacter decreased from 52 to 1 (1.71 log₁₀ reduction; P = .25; Table 1).

A greater than 1 log₁₀ reduction was observed in cultures obtained from sites in both direct and indirect line of site for total CFUs of any organism, total CFUs of the 3 targeted organisms, VRE CFUs, and Acinetobacter CFUs (Table 1). A 0.80 log₁₀ reduction was observed for C. difficile in direct line of sight, but a 1.18 log₁₀ reduction was observed in indirect line of sight. No statistically significant differences were observed in the reductions that occurred in direct versus indirect line of sight disinfection for any of these categories.

The total number of CFUs was reduced following use of UV-C light at each of the 9 environmental locations tested for total CFUs of any organism and each of the 3 targeted organism (Table 2); the greatest reduction was observed on the overbed table (98%), while the lowest reduction was observed on the bathroom floor adjacent to the toilet (74%). VRE was identified on 49 (11%) of 428 plates before use of the UV-C device and only 6 (1%) of 428 plates afterward (P < .0001; Figure 1). The proportion of C. difficile–positive plates similarly decreased from 10% to 5% (P = .03). While the proportion of Acinetobacter-positive plates decreased from 13% to 3%, this trend was not statistically significant (P = .38).

Numerous studies have demonstrated that current strategies for terminal room disinfection are inadequate. In fact, 50% or more hospital surfaces may go untouched and uncleaned following terminal room disinfection.¹² UV-C light is a novel method to enhance terminal disinfection of hospital rooms. Our multicenter, prospective study confirms that automated UV-C emitters substantially decrease the bioburden of important pathogens, such as VRE and C. difficile, from patient rooms in real-world settings. In addition, our data suggest that an automated UV-C emitter can help reduce the bio-burden of Acinetobacter spp. It was likely that the reduction in the bioburden of Acinetobacter spp. was not statistically significant because of the low frequency of Acinetobacter infection in our study hospitals.

Authors of several studies have investigated the efficacy of automated UV-C emitters against important pathogens in nonclinical, experimental conditions.^15–17 For example, Boyce et al¹⁸ used a quantitative disk carrier method to evaluate the efficacy of UV-C light emitters in reducing the burden of C. difficile spores in patient rooms. Disks inoculated with 10⁵ to 10⁶ nontoxigenic C. difficile spores were placed in specific locations in 25 patient rooms and then exposed to light from an automated UV-C-emitting device. C. difficile spores were reduced between 1.4 and 2.9 log₁₀.¹⁸

To our knowledge, however, only 2 previously published studies have evaluated the effectiveness of automated UV-C emitters in clinical conditions following patient discharge. Rutala et al¹⁹ cultured 10 targeted environmental surfaces in 8 rooms of patients previously placed in contact precautions because of colonization or infection with MRSA. An automated UV-C-emitting device was used before cleaning. UV-C irradiation led to decreases in total CFUs per culture plate, in the number of samples positive of MRSA, and in MRSA counts per plate.¹⁹ Nerandzic et al²¹ performed a similar experiment in 66 rooms of patients previously placed in contact precautions for MRSA or C. difficile. The proportion of sites positive for MRSA or C. difficile significantly decreased following UV-C irradiation. While the proportion of surfaces positive for VRE decreased from 2.7% to 0.4%, this trend was not statistically significant (P = .07).²¹

Our multicenter study utilized similar methods and confirmed that application of measured doses of UV-C light produced greater than 1 log₁₀ reductions in the bioburden of C. difficile spores. These reductions also were observed from environmental sites in both direct and indirect line of sight locations relative to the portable UV-C-light-emitting device. Unlike the study by Nerandzic et al,²¹ we also targeted and sampled patient rooms from which a patient on contact precautions for VRE had been discharged. In this setting, UV-C light led to a 1.68 log₁₀ reduction in colony counts of VRE at sampled environmental sites.

We are not aware of prior studies that have evaluated the efficacy of automated UV-C emitters in reducing the environmental bioburden of Acinetobacter in clinical settings such as patient rooms. In our study, UV-C light led to a 1.71 log₁₀ reduction in the environmental bioburden of Acinetobacter. Similarly, the proportion of sampled locations positive for Acinetobacter decreased from 13% to 1%. Unfortunately, we were able to enroll only 2 rooms for Acinetobacter that met enrollment criteria during our study. Thus, our analysis was statistically underpowered.

Our results and the results cited above demonstrate that UV-C is less effective at killing bacteria in clinical settings compared with experimental, nonclinical conditions. For example, the application of UV-C light via automated devices has been shown to decrease the bioburden of MRSA, VRE, C. difficile, and Acinetobacter by 3–4 log₁₀ in experimental conditions (eg, inoculated formica sheets or discs).¹⁹ In contrast, the use of an automated UV-C light emitter produced a 1.07 log₁₀ reduction in total CFUs and a 1.35 log₁₀ reduction in targeted pathogens in our study. Because cultures were obtained before and after application of UV-C light in rooms that had not undergone standard cleaning and disinfection, it is possible that the efficacy of UV-C was adversely affected by the presence of dirt and debris on surfaces and equipment. For example, the authors of a recent study of the efficacy of hand-held UV-C devices reported that these devices were ineffective for the disinfection of 72% of the 68 computer keyboards located in hospital wards if mechanical cleaning was not performed before attempted disinfection with UV-C light.²² In another study, UV-C was effective in disinfecting ultrasound probes only if it was applied after surfaces were disinfected with mechanical friction and a chemical disinfectant.²³ Nerandzic et al²¹ previously investigated the effectiveness of an automated UV-C emitter following cleaning by environmental staff. No samples from 26 rooms were contaminated with MRSA following both terminal room disinfection and use of the automated UV-C emitter.²¹ These studies and our results suggest that UV-C disinfection may be more effective when used after traditional cleaning protocols. Thus, cleaning must remain an important part of terminal room disinfection, but in its absence or on locations missed by cleaning staff, more than 90% of pathogenic bacteria will still be killed when an automated UV-C device is used.

Our study has limitations. First, we utilized a convenience sample of patient rooms from 2 acute care tertiary care hospitals. Thus, our data may not be generalizable to other settings. Second, the minority of our samples identified pathogens of interest. While this may have limited some of our statistical power, our findings are actually consistent with other published studies. That is, in general, approximately 10%–20% of surfaces are typically contaminated with pathogenic bacteria.^19,21 Third, we were able to study only 2 rooms of patients infected with Acinetobacter spp., thus limiting the statistical power of our results. Finally, our study did not provide any patient-specific information, thereby limiting our ability to make conclusions about the impact of this technology for specific patient groups (eg, immunocompromised patients).

The use of UV-C radiation in medical settings is expanding to novel settings and through novel methods. For example, UV-C has recently been used to decontaminate specific rooms in long-term care facilities²⁴ and has been trialed with hand-held devices.^22,25 Our data support and expand on previously published studies to confirm that automated UV-C-emitting devices can decrease the bioburden of important pathogens in hospital rooms. Whether this method actually leads to improved patient safety, decreased acquisition of pathogenic bacteria, and decreased rates of health care–associated infections remains to be seen.