Flexible gastrointestinal endoscopes are semi-critical devices that contact mucous membranes and can be contaminated with blood, intestinal contents, and other biological material during routine use. Reliable, reproducible reprocessing is central to infection prevention. Real-world microbiological performance depends on the entire reprocessing chain, including bedside pre-cleaning, manual cleaning, automated endoscope reprocessor (AER) cycle performance, drying, storage, and sampling sensitivity. Microbiological surveillance programs use quantitative culture thresholds and predefined indicator organisms to detect residual bioburden as a quality-assurance measure rather than to estimate patient-level infection incidence.
This exploratory study aimed to compare microbiological surveillance outcomes after automated reprocessing of patient-used gastrointestinal endoscopes using three AER systems under routine clinical workflow conditions: a peracetic acid (PAA)-based system and two functional-water systems (ozonated water and strongly acidic electrolyzed water). The objective was to provide transparent, endoscope-level surveillance data to inform quality programs and future studies, not to demonstrate clinical infection prevention or device interchangeability.
A single-center, prospective randomized microbiological surveillance study was performed at a tertiary-care endoscopy center. The Institutional Review Board approved the protocol, and enrollment began on November 8, 2024. The study was registered retrospectively in the University Hospital Medical Information Network Clinical Trials Registry (UMIN-CTR; UMIN000061627) on May 19, 2026. Consent was obtained at bedside by oral explanation and documented in the electronic medical record per IRB-approved procedures.
The unit of allocation and analysis was the endoscope-use episode. Allocation to one of three AER systems was performed 1:1:1 using computer-generated random numbers (RAND in Microsoft Excel) applied to the scheduled examination list on the preceding day.
Consecutive patient-used forward-viewing upper and lower gastrointestinal endoscopes were eligible. After each examination, standardized bedside pre-cleaning was performed immediately, followed by manual cleaning with detergent flushing and channel brushing according to guidance for flexible gastrointestinal endoscope reprocessing. Each eligible endoscope then underwent automated reprocessing with the assigned AER.
The three AERs were:
Each AER was operated under the manufacturer’s validated cycle conditions without experimental modification. Personnel performing reprocessing and sample collection were not blinded; the external microbiology laboratory performed testing without knowledge of allocation.
Sampling occurred at two time points: immediately after completion of the clinical examination before bedside pre-cleaning, and immediately after completion of the AER cycle. Post-reprocessing samples were obtained before alcohol flushing, drying, or storage.
Sampling sites and methods were predefined: for surface sampling an insertion portion swab (site D) or control section swab (site E) was wiped 10 times with a saline-moistened swab; for channel sampling (site F), 20 mL sterile saline was flushed through the biopsy port and recovered at the distal outlet. No neutralizing agent was used in sampling or transport.
Samples were processed by an external laboratory using quantitative aerobic culture on soybean-casein digest agar incubated at 35 °C for 48 hours. Results were reported as colony-forming units (CFU) per sample; organisms were identified by 16S rDNA sequence analysis when applicable. Laboratory reports sometimes used categorical results (e.g., below detection limit or <20 CFU); for endpoint classification these were treated as below the predefined CFU failure threshold.
The primary outcome was microbiological surveillance success for each endoscope-use episode after automated reprocessing. Success was defined as both (1) residual bacterial load ≤20 CFU per sample at every sampled site and (2) absence of predefined indicator organisms after reprocessing. Indicator organisms were prespecified as Escherichia coli, enterococci, Enterobacteriaceae, Pseudomonas aeruginosa and other gram-negative nonfermenters, Staphylococcus aureus, and Staphylococcus epidermidis. Detection of any indicator organism constituted surveillance failure regardless of CFU count.
Secondary outcomes included distribution of CFU counts, failure components, and distribution of failures by endoscope type, sampling method, sampled site, procedural factors, individual endoscope identifier, and repair history.
Because robust preliminary data were unavailable, the sample size (74 endoscope-use episodes per group, total 222) was determined by feasibility during the predefined study period rather than formal power calculations; the study was therefore not powered for a definitive non-inferiority conclusion. The protocol specified non-inferiority comparisons versus the PAA-based system but did not define a numerical margin; the present analyses used a post hoc −5 percentage-point margin and an exploratory −10 percentage-point margin.
Surveillance success rates were reported as proportions. Comparisons with the PAA group were expressed as risk differences with 95% confidence intervals (Newcombe method) and Fisher’s exact test for descriptive purposes. Post hoc sensitivity analyses addressed repeated use of the same physical endoscope using a cluster bootstrap, generalized estimating equations clustered by physical endoscope, and an analysis excluding a single problematic scope (SCOPE005).
A total of 222 eligible endoscope-use episodes were randomized and completed assigned reprocessing and sampling: 74 episodes per AER group. The distribution of upper and lower GI procedures was comparable across groups. Demographic and procedural summaries reported mean ages around the early 60s and similar proportions of diagnostic-only versus biopsy or therapeutic procedures across groups. A prospective screening log of excluded episodes before randomization was not maintained, so numbers screened and excluded could not be reconstructed.
Surveillance success after automated reprocessing was achieved in:
Risk differences versus PAA were −5.4 percentage points (95% CI −14.6 to 3.2) for strongly acidic electrolyzed water and 0.0 percentage points (95% CI −7.7 to 7.7) for ozonated water.
The manuscript reports exploratory descriptive summaries of CFU distributions, failure components, and associations with endoscope type and individual scope identifiers. A provided supporting table lists all 13 protocol-defined surveillance-failure episodes.
Using a post hoc non-inferiority margin of −5 percentage points, non-inferiority was not demonstrated for either functional-water system versus the PAA-based system. When a −10 percentage-point margin was examined as an exploratory sensitivity analysis, ozonated water met that threshold but strongly acidic electrolyzed water did not.
Post hoc scope-level sensitivity analyses produced directionally consistent results with the primary analysis and highlighted that repeated failures concentrated in a single physical endoscope materially influenced overall group findings. These analyses were explicitly labeled exploratory and hypothesis-generating.
This pragmatic randomized surveillance study provides endoscope-level microbiological quality-assurance data comparing a PAA-based AER with two functional-water AER systems under routine workflow conditions. Overall surveillance success was high across groups, with PAA and ozonated water both achieving 95.9% success and strongly acidic electrolyzed water 90.5%.
Key limitations: the sample size was chosen by feasibility and the study was not powered for definitive non-inferiority conclusions; the non-inferiority margin used was post hoc; allocation and sampling personnel were not blinded; the trial registration was retrospective; and repeated failures in a single scope affected results. The authors emphasize that the findings are exploratory microbiological surveillance data and should not be interpreted as direct evidence of clinical infection risk or of comparative clinical effectiveness.
These data can inform local quality-management programs and design of future confirmatory studies but do not substitute for broader clinical outcome evidence or larger, adequately powered non-inferiority trials.