Marc recently posted about the second clinical outcome findings from the BETR-D study, recently published in the Lancet Infectious Diseases. Marc contended that the team may have been ‘blinded by the [UV] light’ in reaching the conclusion that enhanced terminal room disinfection led to a hospital wide reduction in acquisition of key pathogens. Here, in the spirit of healthy academic debate, I offer another perspective.
The first clinical outcomes from the BETR-D study, a multicentre cluster-randomised study of various terminal room disinfection strategies (some of which include UV light) were published in the Lancet last year. The study involved a randomised sequence of four terminal disinfection strategies:
- Reference: QAC (except for C difficile, for which chlorine was used)
- QAC + UVC (except for C difficile, for which chlorine + UVC was used)
- Chlorine + UVC
This study found that there was an overall reduction in target pathogen acquisition rate when patients were admitted to a room from which a patient with the target pathogen has just been discharged (called a ‘seed’ room). This reduction seemed to be driven by MRSA and VRE; there was no measurable impact against C. difficile. This left me concluding that it’s time to go shopping for a UV system! This follow-up study evaluated whether improvements in terminal disinfection would translate into a hospital-wide reduction in the rate of acquisition of target pathogens. Worth noting that because the followup study included all admissions, rather than just those exposed to a potentially seeded room, it included a whopping 271,000 patients (almost 10-fold more than the 30,000 in the BETR-D study part 1)!
The primary outcome was a composite acquisition rate of the target pathogens: MRSA, VRE, C. difficile, and multidrug-resistant Acinetobacter. Now, as Marc pointed out, this primary outcome measure was not met because although there was a reduction, this was not statistically significant. However, looking at the numbers, it was as close to statistically significant as you can be, with the upper confidence interval bang on 1: for UVC vs. reference relative risk (RR) = 0.89, 95% CI 0.79–1.00; p=0.052. Whilst I accept totally that not statistically significant is not statistically significant, we know that the set-point for statistical significance is arbitrary (although widely used), so a finding that is not statistically significant can still be clinically significant. Put another way, would the interpretation of the study have been any different if the p value was 0.048 rather than 0.052?
Also worth noting that there were statistically significant hospital-wide reductions in the acquisition of VRE (RR=0.56, 0.31–0.996; p=0.048) and C. difficile (RR=0.89, 95% CI 0.80–0.99; p=0·031) during the UVC period. You may argue that these findings are not biologically plausible: how can an improvement in terminal disinfection possibly impact hospital-wide acquisition rates when direct room-to-room transmission almost certainly only explains a small proportion of transmission? Well, in a study with a similar design, we tested the impact of improving terminal disinfection of individual ‘seed’ rooms (in this case using HPV) on the ward-scale rate of environmental contamination. There was a reduction in the proportion of rooms contaminated with any MDRO, and especially multiple MDROs and MDROs that differed from the current room occupant, when HPV was in operation.
The hospital-wide reduction in VRE is in line with the findings of the first BETR-D study. However, the hospital-wide reduction in C. difficile acquisition is a surprise, because this did not come out of the first BETR-D study. The design of the BETR-D study meant that rooms known to contain a patient with C. difficile were disinfected with chlorine in all arms of the study, and the measured compliance with the cleaning process was extremely high. So, it’s perhaps not surprising that UVC disinfection showed no incremental benefit over well-performed chlorine disinfection. What could drive a hospital-wide reduction in C. difficle acquisition, however, is that there’s a huge population of patients with asymptomatic C. difficile shedding spores all over the place. To put this into context, in one US study, 12% of patients were asymptomatic carriers of toxigenic C. difficile, and environmental contamination was detected in 17% of their rooms. So, a good number of rooms targeted for UVC disinfection for other pathogens would also be contaminated with spores from asymptomatic carriers. And dealing with these spores more effectively using UV could drive a hospital-wide reduction in C. difficile.
One interesting finding of the study was that only 30% of the uses of UVC were indicated according to the study protocol i.e. for a known ‘seed’ room. The team evaluated the impact of both indicated and these ‘non-indicated’ uses of UVC by comparing patient outcomes in rooms disinfected using UVC for any reason. The findings were similar in shape to the findings from the original BETR-D study, showing a reduction in rate of patient-level acquisition (although no stats were run because this was a post-hoc analysis).
Overall, this is a huge and very impressive study. Taking the findings of both the first and the second BETR-D studies together, I think there is compelling evidence that introducing UVC to a terminal disinfection protocol improves patient outcomes both for the patients admitted to the room that is disinfected using UV, and for other patients across the hospital.