Mairead Smith wheeled the beige, 124-pound contraption into place near a hospital bed, pushed the start button, and quickly got out of the room.
Within moments, peering through a viewing window, she could see a ghostly, bluish light.
Such devices are designed to kill germs that have survived the efforts of hospital cleaning staff, by irradiating the room with invisible ultraviolet light. (The accompanying blue light just looks cool, and also serves as a reminder for people to stay away. Long-term exposure raises the risk of cancer; in the short term, the light can cause eye damage called welder's burn.)
Studies have found that the devices, commonly called robots despite the lack of automated moving parts, seem to reduce the rates of hospital-acquired infection. But there has been little independent study comparing the various brands, said Smith, an engineer at ECRI Institute in Plymouth Meeting.
Are they all equally good at killing bacteria, particularly under bed railings and in other shadowy nooks that lie outside the direct path of the machine's UV rays?
Smith and her colleagues at ECRI, a nonprofit medical research organization, are on the case.
They have tested half a dozen models, stationing them at several places in a mock hospital room. Using small, cylindrical sensors, they measured how much energy struck 11 different spots around the room, with a particular emphasis on "high-touch" areas.
ECRI scientists also visited an external microbiology lab to test how much UV radiation was needed to kill one of the baddest bugs of them all: Clostridium difficile.
An infection with C. diff, as it is known for short, can cause severe diarrhea and excruciating abdominal pain. Within a month, it is fatal in one out of 11 hospital patients over 65, according to the federal Centers for Disease Control and Prevention.
The ECRI analysis, which is not yet complete, is focusing on this microbe for a simple reason, Smith said.
"If you can kill C. diff, you've probably killed everything else," she said.
The bacteria give hospital officials fits because for part of their life cycle, the organisms take the form of hard-to-kill, encapsulated spores that can live for months. Bleach will do the trick, unless you miss a spot. That is where UV light comes in, but it is not cheap. The machines can run into the low six figures.
The 124-pound model Smith was testing, made by UVC Cleaning Systems Inc. of Marlette, Mich., has a sticker price of more than $55,000 but generally sells for less as part of package deals, company owner Robert Gilling said.
Another option is a device that sprays the room with a hydrogen peroxide mist, which is great for penetrating nooks and crannies, said Smith's ECRI colleague, engineering manager Jeremy Suggs.
"It fills the rooms," he said. "There aren't any shadows."
But the turnaround time is longer. Typically the mist treatment requires a couple of hours, including the time for the substance to dissipate, whereas a room disinfected with UV light is usable as soon as the machine is turned off — after letting it run for half an hour or so.
Even when hospitals employ multiple weapons, C. diff infections persist.
In a February study in the Lancet, led by Duke University's Deverick J. Anderson, the rate of C. diff infection in rooms treated with bleach and UV light was no different from the rate in rooms cleaned with bleach alone.
That lack of difference may have resulted from good bleach technique, said Anderson, an infectious disease specialist with Duke's health system. Cleaning staff used bleach on 90 percent of surfaces in the study, compared with 50 percent in some previous research, he and his coauthors wrote.
"If we do pretty well with the bleach, there may not be a whole lot left for the UV light to help," Anderson said.
Also, in previous studies that found UV light could help against C. diff, researchers had looked at hospital-wide infection rates, whereas Anderson and colleagues focused on the risk for people in rooms that were previously occupied by patients with known infections.
Whatever the cleaning technique, hospitals in Anderson's study and elsewhere seem unable to get to zero. Nationally, rates seldom fall below half a dozen cases per 10,000 patient days.
That may be due to patients who enter the hospital with the bacteria already colonized inside them, Anderson said. Then, if they are treated with antibiotics to ward off some other kind of infection, that leaves a clear playing field for a C. diff infection to flourish, he said.
Told of the ECRI study, he said:
"That will be very useful information in the big picture. It seems that C. diff is ultimately more complicated than our initial or perhaps even current theories really allow for."
Area facilities with UV light machines include Doylestown Hospital, which acquired a Xenex brand device in 2013. The hospital has added other strategies, as well, such as enhanced surveillance of cleaning staff, spokesman Ron Watson said.
The hospital's C. diff infection rate has not changed much since then, remaining in the single digits of cases per 10,000 patient-days, statistically equivalent to a nationally established benchmark.
Would more ultraviolet light devices help?