Feedback of ATP Measurement as a Tool for Reducing Environmental Contamination in Hospitals in the Dutch/Belgian Border Area.

Background & objective: The objective of this study was to determine the effect from feedback of ATP measurements on environmental contamination within hospitals in the Dutch/Belgian border area. Setting: Standardized ATP measurements were conducted in 9 hospitals on pre-dened fomites. Four different fomite groups were dened: medical devices, patient bound materials, ward bound materials and sanitary items. ATP results were reported in relative light units (RLU), RLU >1000 was considered as “not clean.” Two rounds of ATP measurements were conducted. After the rst round of ATP measurements, results were provided to the wards. There were no structured cleaning interventions implemented. The second round of ATP measurements was performed one year later. The amount of surface contamination before and after the feedback were compared. Results: In total 1923 ATP measurements were performed. Before feedback 960 ATP measurements were conducted and after feedback 963. The overall median reduction in RLU was 381 (p<0.001), from 568 before feedback to 187 afterwards. In each hospital there was a reduction of the median RLU after feedback. Conclusion/discussion: Substantial reductions in RLU values were found after feedback of ATP measurements. Feedback of ATP measurements was associated with a major reduction of surface contamination in hospitals.

Results: In total 1923 ATP measurements were performed. Before feedback 960 ATP measurements were conducted and after feedback 963. The overall median reduction in RLU was 381 (p<0.001), from 568 before feedback to 187 afterwards. In each hospital there was a reduction of the median RLU after feedback.
Conclusion/discussion: Substantial reductions in RLU values were found after feedback of ATP measurements. Feedback of ATP measurements was associated with a major reduction of surface contamination in hospitals.

Setting
As part of a multicenter project in the Dutch/Belgian border area, the i-4-1-Health project, standardized ATP measurements were conducted in 9 hospitals (3 Belgian university hospitals, 1 Dutch university hospital, 3 Dutch teaching hospitals and 2 Dutch general hospitals). ATP measurements were conducted on different hospital wards, ranging from 2 wards up to 4 wards per hospital, depending on the hospital size. In each hospital, ATP measurements were conducted on a surgical ward, an internal medicine ward and if applicable two other medical wards. When ATP measurements were conducted on more than 2 wards a selection was made from the medical specialties urology, cardiology, orthopedic surgery, pulmonology and/or geriatrics. For the data analysis medical specialties were merged into two groups: surgical specialties and non-surgical specialties. On each ward, ATP measurements were performed on a selection of 30 pre-de ned fomites. 9 These fomites were classi ed into 4 different groups: medical devices, patient bound materials, sanitary items and ward bound materials. Fomites were chosen based on the following criteria: frequently touched by nursing staff or patients or being in the direct vicinity of patients or high risk surfaces (e.g. tabletop for medication preparation). ATP measurements were performed at two points in time, one year apart each other. After the rst round of ATP measurements feedback was given to the nursing and cleaning staff. The ATP results were provided to the ward by an infection control practitioner.
There were no structured cleaning interventions planned as part of the project. Each hospital was free to implement any cleaning interventions. The effect of the feedback was measured in a second round of ATP measurements.

Atp Measurements & Rlu Breakpoints
The Clean-Trace NG Luminometer (3M, Zoeterwoude, the Netherlands) was used for the ATP measurements, results were reported in RLU. ATP measurements were conducted by trained and validated researchers working at the department of infection control of the corresponding hospital. The RLU < 1000 breakpoint for cleanliness was de ned for measuring a fomite at a random point during the day as described in van Arkel, el al. 9 An RLU value above 1000 RLU was categorized as unclean or intermediate, above 3000 RLU as dirty.

Statistical Methods
All data were analyzed with Statistical Package for Social Science software (SPSS; IBM Corp., Armonk, New York, US; version 25) and R (R Foundation, New Zealand, R version 3.6.2). Adjusted relative risks (ARRs) were calculated based on the differences in the occurrence of "not clean" fomites (RLU > 1000) between the two time periods and analysed using mixed effects Poisson regression models using a log link with a random intercept and random slope per hospital. Differences between RLU values were analysed log transformed using mixed effect linear regression models with a random intercept and xed slope per hospital. Adjusted models were corrected for hospital, medical specialty and surface category.

Results
In total 1923 ATP measurements were performed. Before feedback 960 ATP measurements were conducted and after feedback 963. Per hospital 120 up to 246 ATP measurements were performed, depending on the hospital size.
The median RLU before feedback was 568 RLU and after feedback 187 RLU, resulting in a reduction of 381 RLU (p < 0.001) (Fig. 1). Of all measurements before feedback 37.7% (362/960) were considered as "not clean" (RLU > 1000), after feedback 13.1% (126/963) were considered as "not clean." The differences in RLU between the rst and second round per hospital are visualized in Fig. 2. The median RLU value per hospital before feedback ranged from 279 to 2137 (Fig. 2). After feedback the median RLU value per hospital ranged from 83 to 830 (Fig. 2). Each hospital showed a reduction in median RLU between the rst and second round of measurements.
Per medical specialty between 60 and 538 ATP measurements were conducted. The median RLU value before feedback was 627 in the surgical specialty group and 546 in the non-surgical specialty group. After feedback the median RLU value was 200 in the surgical specialty group and 172 in the non-surgical specialty group.
Per fomite group 320 up to 640 ATP measurements were conducted: 627 ATP measurements in the medical devices group, 320 ATP measurements in the patient bound materials group, 640 ATP measurements in the sanitary items group and 336 ATP measurements in the ward bound materials group. The differences in RLU between rounds of the fomite groups is visualized in Fig. 3. The median RLU value in the patient bound materials group was reduced from 931 to 224, in the ward bound materials group from 659 to 293, in the medical devices from 651 to 187 and in the sanitary items from 396 to 131, before and after feedback respectively.
Predictors for the more frequent occurrence of "not clean" (RLU > 1000) surfaces between the rst and second round of measurements for the different groups are visualized in Table 1, signi cant differences are highlighted. Table 1 Univariable and multivariable analysis of median differences and percentages of "not clean" (RLU > 1000) items per round, with adjusted relative risks. Signi cant differences in bold (p < 0.05). Adjusted models are corrected for hospital, medical specialty and surface category. The results of the ATP measurements increased interest and motivation for cleaning amongst the nursing and cleaning staff. We observed repeatedly that nursing staff voluntary measured different surfaces on the ward to get insight into surface contamination and consequently improved cleaning of these surfaces.
Previous research has shown that performing ATP measurements has a bene cial effect on cleaning on hospital wards, by having an effect on multiple factors e.g. motivation of hospital staff for cleaning surfaces, giving insight into contamination of different surfaces/groups of fomites and giving a quanti able outcome of measurement. 6,7 Feedback of ATP measurements to nursing and cleaning staff seems to be an effective method to improve cleaning of hospital wards. Moreover, previous research has shown that cleaning can be improved by implementing relatively simple changes in the cleaning protocol. 1 Within this study each hospital was free to implement cleaning interventions. These interventions included de ning cleaning responsibilities per fomite, educational sessions for cleaning staff and/or introduction of new cleaningwipes. In some hospitals it was indicated that there were no cleaning interventions implemented.
For this study RLU thresholds were copied from a previous study. 10 Different studies have recommended an RLU threshold for cleanliness at 250-500 RLU, however this threshold is intended for measurement (almost) directly after cleaning. 1,3,11,12,13,14 Within this study RLU thresholds for conducting ATP measurements at a random point in time were used. The goal of this study was to improve cleaning based on feedback from ATP measurements. By using above described thresholds feedback could be given in an easy to visualize way.
During this study, important improvements in hospital cleanliness were observed. Considering the size of the improvement and that it was observed in all centers, it is plausible that these improvements can be contributed to feedback from the ATP measurements. However, it is unknown how long this effect will be maintained. To obtain a sustainable effect, repeated measurements over time will probably be needed.
Indeed, other studies have found a washout effect after ATP measurement was ceased. 15 In general, a quality program is characterized by repeated measurements and subsequent actions to improve the result.
As such, ATP measurements should be integrated in a quality system for environmental cleaning.
Within three hospitals there was an outbreak with vancomycin-resistant enterococci (VRE) between the rst and second round of ATP measurements. Consequently, there was a better focus on cleaning on the affected hospital wards. A part of the decrease in RLU values could be explained by the cleaning measures implemented during these VRE outbreaks. However, a signi cant decrease in environmental contamination was found in almost all hospitals, indicating that feedback from ATP measurements still has a bene cial effect on hospital cleanliness in general.

Conclusion
Substantial differences in RLU values were found after feedback of ATP measurements. The second round of measurements showed signi cantly lower median RLU values in all groups (hospitals, surface categories and medical specialties), together with signi cantly lower percentages for "not clean" surfaces (RLU > 1000) in all groups. These ndings suggest that feedback of ATP measurements, presented in a way that is easy to understand, has a bene cial effect on cleaning in general. Furthermore, ATP measurements give insight into speci c areas with a high level of environmental contamination to guide speci c interventions.

Declarations
Ethics approval and consent to participate Not applicable.

Consent for publication
Not applicable.

Availability of data and materials
As agreed within the i-4-1-Health consortium, the i-4-1-Health datasets will be made available no earlier than December 31st, 2020 and no later than December 31st, 2024, in accordance with the FAIR (Findable, Accessible, Interoperable and Reusable) data principles. 16

Competing interests
The authors declare that they have no competing interests.

Funding
The i-4-1-Health project was nanced by the Interreg V Flanders-The Netherlands program, the cross-border cooperation program with nancial support from the European Regional Development Fund (ERDF).
Additional nancial support was received from the Dutch Ministry of Health, Welfare and Sport, the Dutch