https://orcid.org/0000-0003-0643-2210
Figures
Abstract
Introduction
Despite increasing evidence that Environmental Health Monitoring (EHM) is an effective method to perform rodent colony health surveillance, promotes the 3Rs as a Replacement, is comparable or reduces cost, and demonstrates labor benefits, many research institutions continue to use live Soiled Bedding Sentinels (SBS). Some veterinarians and others responsible for rodent colony health monitoring remain cautious of the evidence supporting this new technology. Therefore, our objective was to perform a formal systematic review to identify, summarize, and evaluate the research on the efficacy of EHM as compared to SBS. This information is essential to establishing recommendations for rodent health monitoring programs.
Methods
We systematically searched and evaluated all articles comparing EHM to SBS identified from PubMed, Web of Science, and CAB abstracts initially on November 12, 2021 with an update on Oct 15, 2023. Our inclusion criteria included publication in a peer-reviewed journal and collection of empirical data on rodent health monitoring alternatives to soiled bedding sentinels. Outcome data was extracted and analyzed via logistic regression and secondary descriptive statistics. Due to the nature of the included studies, no risk of bias assessment was performed. No specific funding was received for this review.
Results
Forty-two peer-reviewed articles passed inclusion criteria for analysis. The design of studies varied substantially but included publications investigating exhaust dust testing (n = 27), sentinel-free soiled bedding (n = 8), and direct colony sampling (n = 24). Based on both logistical regression and descriptive criteria, all types of EHM appear to be superior to soiled bedding sentinels at detecting pathogens, regardless of their taxonomic classification.
Conclusion
Based on these findings, we conclude there is a strong evidence base supporting the replacement of SBS with EHM. Looking forward, we encourage continued research on the detection of existing and emerging pathogens with these technologies and development of best practices for routine rodent health monitoring programs.
Citation: LaFollette MR, Clement CS, Luchins KR, Manuel CA, Foley PL, Hanson WH, et al. (2024) Do we still need a canary in the coal mine for laboratory animal facilities? A systematic review of environmental health monitoring versus soiled bedding sentinels. PLoS ONE 19(12): e0311840. https://doi.org/10.1371/journal.pone.0311840
Editor: I. Anna S. Olsson, Universidade do Porto Instituto de Biologia Molecular e Celular, PORTUGAL
Received: February 12, 2024; Accepted: September 25, 2024; Published: December 5, 2024
Copyright: © 2024 LaFollette et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the manuscript and its Supporting Information files.
Funding: The author(s) received no specific funding for this work.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Rodent health monitoring programs detect unwanted pathogens in animal research facilities. As such they help ensure rodent health, personnel health, and the reproducibility and reliability of scientific research. Traditionally, rodent health monitoring programs have relied on sentinel rodents that are not directly involved with experimental studies. Rodent sentinels are exposed to pooled, soiled bedding from research animals in the colony. The rationale behind this is that the soiled bedding sentinel (SBS) animal becomes infected through exposure to a pathogen originating from the research colony. The disease is then detected in the sentinel animal by a variety of diagnostic assays or clinical signs. Typically, sentinel rodents have an intact immune system for pathogen detection by serology, but immunodeficient strains have also been used to better represent the population under surveillance or propagate the pathogen for easier detection by molecular methods. Sentinel animals are typically exposed at each cage-change, serially sampled, euthanized, and then replaced after 3–6 months. There is usually one SBS cage per rack, or per side of a rack, both to give specificity as to the location of a pathogen and to increase sensitivity by avoiding dilution of the pathogen in too large of a pool of soiled bedding. In any given research institution, hundreds or even thousands of rodents may be used each year exclusively for routine health surveillance without directly contributing to research data.
Despite SBS being standard practice for health monitoring programs, a 2016 systematic review of the SBS methodology could find only 15 relevant papers [1]. Systematic reviews follow a formal process to perform an exhaustive unbiased search of the literature, compile the results of individual papers into a meta-dataset, and then analyze this to detect consistent patterns of results [2]. De Bruin et al. [1] were able to find consistent evidence (in at least two studies) for the efficacy of SBS for mouse hepatitis virus (MHV), mouse parvovirus (MPV), Theiler murine encephalomyelitis virus (TMEV), Helicobacter spp., and fur mites; and evidence for a lack of efficacy for Sendai virus. However, for several other pathogens, there were either inconsistent results or evaluation by only a single suitable study [1]. From a cost- benefit and 3Rs perspective, the lack of supporting literature for SBS is concerning.
In recent years, Environmental Health Monitoring (EHM) has emerged as an alternative to SBS based health monitoring [3–10]. The rationale behind EHM is to bypass the use of sentinel animals and detect pathogens in the environment directly by monitoring for the presence of their genetic material. In this context, the term “environmental” refers to space in the macro-environment of the housing room or within the mechanics of the caging system or micro-environment of the intra-cage space directly around the animal. Instead of live animals, EHM uses sterile swabs, flat media, or filter material to collect diagnostic samples within the animal cage, husbandry equipment, or soiled bedding. From these samples, the presence of pathogens in the environment is then detected by sensitive and specific PCR assays. A wide variety of different protocols for sample collection exist (see Table 1), but generally, they fall into 4 main approaches: Exhaust Dust Testing (EDT) which collects samples from the primary housing equipment; Sentinel-Free Soiled Bedding (SFSB) which collects samples from pooled soiled bedding; Room and Equipment Monitoring (REM) which collects samples from the room or husbandry equipment; and Direct Colony Sampling (DCS) which collects samples directly from colony animals themselves by non-invasive or minimally invasive methods.
EHM, in theory, has three distinct advantages over SBS. First, it avoids biological complications inherent to detecting the immune response of soiled bedding exposed sentinels as proxy for pathogen presence. More specifically, the mode of transmission for a variety of commonly excluded rodent pathogens is not through the fecal-oral route [11–13], and there are host-adapted and environmentally labile pathogens not readily transmitted by soiled bedding [14]. Furthermore, sentinel susceptibility to infection may vary by animal age and strain [15–20]. Additionally, when testing is reliant on the occurrence of infection and seroconversion, the time to seroconvert can result in significant delays in detection [21]. [14] Second, because the sensitivity of EHM comes from the sensitivity and specificity of PCR based assays, EHM avoids the complexity of using differing methods of detection for different pathogens and can detect even minute levels of pathogen genomic sequences in samples collected from the environment [14]. Third, EHM eliminates any animal welfare concerns as well as removes animal costs and decreases labor costs inherent to a SBS program [22]. Thus, EHM is a potentially powerful 3Rs Replacement and can also benefit facility operations and overhead costs [22, 23].
Despite the advantages of EHM, EHM remains largely unadopted. In 2017, 95% of institutions were using SBS for rodent health monitoring [24]. Still, as of 2021, 89% of surveyed institutions were continuing to use SBS [22]. In the latter survey, 16% of participants believed EHM to have lower accuracy than SBS and 40% believed it to be difficult to use due to their caging types. Thus, a key roadblock to wider implementation is the perception of EHM as an inferior method compared to SBS, despite literature to the contrary.
As illustrated by de Bruin et al.’s work on SBS programs [1], systematic reviews are especially important when the scientific community does not have consensus on evidence-based best practices. First, they summarize the entirety of scientific literature on a particular topic in one place and subsequently expose decision-makers to evidence they may not have otherwise identified. Second, they counter confirmation bias (i.e., the subconscious tendency to pay more attention to evidence supporting a position and to discount evidence that refutes it). Third, they expose methodological issues and establish best practices in a particular research arm for the future.
Accordingly, here we present a systematic review of the evidence-base comparing the efficacy of EHM as compared to SBS. We had 4 aims that we aspired to achieve with this work. First, we formally compared the sensitivity of EHM to SBS across the current literature. Second, we formally tested whether any EHM methodologies were superior across all pathogens. Third, we formally tested whether any methods were better for detecting specific pathogens. Fourth, we provide an exhaustive catalog and qualitative summaries of the existing literature. In brief, this work shows that existing literature overwhelmingly supports EHM.
Materials & methods
Protocol
Prior to conducting the systematic review, we consulted the preferred reporting guidelines for systematic reviews and meta-analysis (PRISMA) guidelines, the Systematic Review Centre for Laboratory Animal Experimentation (SYRCLE) guides, and a veterinary science librarian. A completed PRISMA checklist can be found in S1 File. Our procedures were defined a priori in a study review protocol where we specified the search strategy, inclusion and exclusion criteria, and key data extraction items S2 File. We note that this file was submitted to the SYRCLE database after it had closed so the study cannot be considered formally registered. Additionally, we note the following changes to the protocol: conference proceedings were not searched to prevent duplication, no risk of bias was performed (see below for rationale), multiple reviewers extracted data, and discrepancy was resolved via discussion.
Search strategy and article identification
Articles were identified first on November 12, 2021 and then on October 15, 2023 by searching PubMed, Web of Science, and CAB Abstracts (via CABI). We searched through titles, abstracts, and keywords. Search terms included a term related to environmental health monitoring or exhaust air dust. Language was restricted to English. The full search strategy can be found in S3 File. Search records were uploaded to Covidence, de-duplicated, and screened.
Eligibility criteria and article selection
After all duplicate articles were removed, all articles were initially screened by a single reviewer (CSC) for inclusion based on title and abstract. A random 20% of articles were independently assessed for inclusion by a second reviewer. Any disagreements were resolved via consensus. No automation tools were used. Prior to data extraction, full articles were additionally screened. The following inclusion criteria were used: (1) publication in a peer-reviewed journal; (2) published in English; (3) collection of original data; (4) with laboratory rats or mice; (5) assessing at least one environmental health monitoring technique. Studies were included even if they did not compare the efficacy of EHM to SBS directly. Exclusion criteria included: (1) reviews, (2) other species, (3) no use of EHM, (4) no relevant outcomes measure, and (5) not in English.
Data extraction
Sampling methodology.
For this review and to facilitate conversations in the field, common terminology was created to represent the variety of current methods constituting EHM (Table 1). EHM programs can further be broken down into 4 categories including: exhaust dust testing (EDT), sentinel-free soiled bedding (SFSB), research equipment monitoring (REM), and direct colony sampling (DCS). Several of these categories have been broken down further based on specific sampling methods. For each category, one individual (CSC) with an additional individual confirming (MRL) extracted the primary types of sampling methods that were evaluated in the publication.
In addition to the broad sampling types, one individual (CSC) also extracted potentially relevant information related to sampling methods including sampling material used (Media [including filter paper, filter media, contact media, gauze, etc.], Non-Adhesive Swabs (including cotton swabs, flocked swabs, and dry swabs), and Adhesive Swabs [i.e., Sticky Swabs]), SFSB sampling type (Agitation, Stagnant Exposure, or Dredging), length of time for media exposure, IVC rack sanitation details, total length of monitoring period, type of caging, IVC air changes per hour, frequency of health monitoring, mouse strain and sex, and bedding type. We also used three categories to describe information about EHM sampling material.
Pathogen detection
We extracted which pathogens were tested for and designated whether they were detected with a (yes/no). A pathogen was considered to be detected if the sampling method picked up pathogen nucleic acid at least some of the time (for example 1 of 4 sentinel animals or 25% of media). Two coders per paper (CSC and a veterinarian) determined whether each sampling type detected (at least some of the time) each individual pathogen. Any discrepancies were resolved via discussion.
Pathogens were then recoded in three phases as shown in S2 Table. Coding categories were determined and confirmed by the laboratory animal veterinary co-authors (KRL, CAM, WHH, PLF, CPB, CBW). In the first phase, pathogens were grouped into specific category such as grouping Helicobacter bills, Helicabacter ganmani, Helicobcter spp., and Helicobacter typhlonius into a single Helicobacter spp. group. In the second phase, pathogens were lumped according to specific type such as grouping fur mites, M. musculinus, Myobia musculi, and Radfordia affinis into Ectoparasites. In the third phase, pathogens were grouped via two methods. For method one, pathogens were coded based on type into the following categories: bacteria, fungus, parasite, or virus. For method two, pathogens were coded based on the agent’s relative importance to the lab animal community which was modeled after the FELASA working group on revision of guidelines for health monitoring of rodents and rabbits that details which pathogens are recommended to be tested for on an annual or 3-month basis versus pathogens that are optional to be tested [25].
Research coding
In addition to the above information, one individual (CSC) also extracted data based on the first author, publication year, journal name, country, title, terminology, number of cages, animals replaced, and any notes on money or time saved. Any missing information was left blank during data extraction and subsequently not included in summary statistics.
Reporting quality & risk of bias assessment
Although due to the nature of the experimental designs of these studies, which included studies with no control conditions or animals, the ability to conduct a formal risk of bias assessment was limited, a modified version of the SYRCLE risk of bias assessment for animal studies was conducted by a single researcher (MRL). Results can be found in S2 Table. As neither animals nor cages are typically assigned to different treatment groups, question 1, 3, and 4 were removed from the tool; question 5 was also removed as interventions are not applied to a particular animal or cage. Additionally, for studies that only evaluated a single health monitoring method, were a cost analysis, or case study, all questions were considered not applicable.
Data analysis
Data visualization and final preparation was performed in JMP Pro 16 for Windows and exported to SAS 9.4 for Windows for model selection and analysis. Given the differences in methodology and frequent lack of details regarding sample size, it was not possible to perform a formal weighted meta-analysis [26]. Instead, as described above, for each article for each combination of pathogen and detection method, we scored whether or not the method successfully detected the pathogen. The resulting data were analyzed as a logistic regression using SAS PROC GENMOD to implement a suitable generalized linear model. The dataset used for analysis and SAS code can be found in S1 Dataset. One of the advantages of this approach is that it is robust to missing data. Thus, studies which do not include SBS as a control, or do not compare all EHM methods, or all pathogens, can still be included and contribute appropriately to figuring the average efficacies of each factor combination. Note that data from Eichner and Smith 2023, while included in the descriptive statistics (i.e., total number of articles, pathogens evaluated, EHM methods use, etc.), was not included in formal data analysis due to significant published concerns with the experimental design and data collection [27, 28].
Our initial model was blocked by Article. For hypothesis testing, Sampling Method was nested within EHM vs SBS. This nesting approach implements a Hierarchical Linear Model, which explicitly tests whether SBS differs from EHM on average, and then whether the different EHM methods differ from each other, without the need for additional post-hoc tests. This approach is more powerful than comparing all 4 methods directly and also avoids p-hacking and HARKing by limiting the hypotheses to those defined ahead of time. We ran two versions of this analysis, where pathogen was coded by Type or by Importance (as described above). The interaction of EHM vs SBS by Pathogen Type or Importance (respectively, depending on analysis) was included to test for differences in the efficacy of Sampling Method for different Pathogens.
REML solutions suffer when model fit is too accurate, resulting in complete or quasi-complete separation of the model solution (i.e., the parameter estimates of the underlying regression model). When this happens, the parameter estimates expand with each iteration of the REML algorithm, and never reach a stable solution. This can be due to model overspecification, where unique combinations of variables perfectly predict outcomes or when a hypothesis-testing factor is too powerful in terms of predicting outcomes. We followed best practice in resolving these issues [29]. Overspecification is typically due to having too many categories within different variables. For experimental variables, such as specific pathogen, overspecification was solved by recoding the data into functional categories as described above. For blocking variables (like Article), which cannot be further condensed, these variables can be retained [29].
The initial model showed complete separation and could not generate a stable solution. This was due to including Articles which acted as a blocking factor, as most articles were missing many combinations of Sampling Method and Pathogen Type or Importance. Accordingly, we removed Article from the analysis following best practices for resolving these issues [29]. This yielded a quasi-complete solution with inflated estimates and errors. The interaction of EHM vs SBS by Pathogen Type or Importance was non-significant in either model and was removed. Once removed, the final model yielded a stable solution. The least square means and standard errors generated from the final models were subsequently graphed.
Having identified our final model, we then stress-tested the robustness of our findings by adopting alternative approaches, and trying to “break” the result. This approach is an important prophylactic to p-hacking. It is more recently referred to as “sensitivity analysis” [30] but in reality merely reflects long-standing best-practice in linear model design in whole-animal biostatistics [31]. To do so, we addressed two potential issues with our approach.
First, with respect to removing Article from the model, an alternative approach to the best practice recommended in [29], would be to treat Article as if it was a subject in a repeated measures analysis (i.e. as a random effect). While this approach is common in meta-analysis of continuous outcome variables [2], implementation in a logistic regression is complex. Most importantly, doing so involves Generalized Estimating Equations, which are sensitive to underlying assumptions and implementation, and often do provide consistent results across different software packages [32]. Furthermore, treating Article as a random effect is inherently questionable, not least because random effects pertain to a much larger unsampled population (of articles), whereas a systematic review aims to find the entire population of relevant articles. Despite these reservations, we performed this alternative analysis (see options in S1 Dataset). The results were unchanged, and so we present the more traditional best practice analysis described above.
Second, we initially included all articles, even if they did not include a SBS control, allowing the analysis to correct for these missing data. Accordingly, we reanalyzed the data using only articles which directly compared SBS and EHM, including stress-testing with the General Estimating Equations alternative approach described above. These analyses can be found in S2 Dataset. The results were unchanged, and so we present the analyses from the full data here.
Descriptive analysis.
In addition to the logistic regression performed, we performed several descriptive analyses, tabulation, and visual display to synthesize the results. The datasets used for these analyses can be found in S1 Table which reports a detailed list on pathogens per article and S2 Table which present relevant article level data. These allow for a more granular view of the data on a per pathogen basis that is not possible via logistic regression. We present summarized counts of the number of articles that evaluated each pathogen and the number that successfully detected each pathogen. Additionally, we present counts of various study characteristics.
Results
Study selection
As shown in the PRISMA diagram in Fig 1, our literature search resulted in 862 peer-reviewed publications. After excluding duplicates, a total of 775 abstracts were screened; a full list of these studies with reasons for exclusion can be found in S3 Table. After meeting inclusion criteria based on abstracts, a total of 53 full-text papers were subjected to full-text screening. Ultimately, based on our inclusion criteria, the final sample resulted in 42 articles (5.42% of the total initial pool after duplicates were removed) published between 2004 and 2023. There was an international representation of researchers including corresponding authors from North America (n = 27), Europe (n = 9), Asia (n = 5), and South America (n = 1). The articles were published primarily in the Journal of the American Association for Laboratory Animal Science (n = 23).
Efficacy of sampling method and pathogen
The extracted data were analyzed by Pathogen recoded as Pathogen Type or Pathogen Importance. When Pathogens were coded by Pathogen Type, the Pathogen Type as it related to Sampling Method interaction was non-significant and caused quasi-complete separation of the model and was thus removed. Across sampling methods, detection rate did not differ by Pathogen Type (LR Chi-Sq = 3.34; P = 0.3422; Fig 2). Detection rate was significantly higher for EHM methods overall vs SBS (LR Chi-Sq = 18.60; P<0.0001; Fig 2). The EHM methods did not differ significantly amongst themselves (LR Chi-Sq = 2.04; P = 0.3597; Fig 2).
The detection rate (LSM +/- SE) of A) each rodent health monitoring strategy (soiled bedding sampling (SBS), exhaust dust testing (EDT), sentinel-free soiled bedding (SFSB), or direct colony sampling (DCS)) across B) pathogens coded by type (bacteria/fungus, ectoparasites, endoparasites, viruses) as investigated in 30 articles. N = 252, ***p < 0.0001.
When Pathogens were coded by Importance, the Pathogen Importance by Sampling Method interaction was non-significant and caused quasi-complete separation of the model and was thus removed. Across sampling methods, detection rate was higher for pathogens tested on a 3-Month or Annual basis as compared to Optional tested (LR Chi-Sq = 5.33; P = 0.0210; Fig 3). Detection rate was significantly higher for EHM methods overall vs SBS (LR Chi-Sq = 18.90; P<0.0001; Fig 3). The EHM methods did not differ significantly amongst themselves (LR Chi-Sq = 2.18; P = 0.3358; Fig 3).
The detection rate (LSM +/- SE) of A) each rodent health monitoring strategy (soiled bedding sampling (SBS), exhaust dust testing (EDT), sentinel-free soiled bedding (SFSB), or direct colony sampling (DCS)) across B) pathogens coded by “importance” (per FELASA guidelines as to whether they should be tested on a 3-month/annual basis or optionally) as investigated in 30 articles. N = 252, **p < 0.001 *p<0.025.
Pathogen-level descriptive analysis
In addition to a formal statistical analysis, we also present the data on a per pathogen basis which is summarized in Table 2 with detailed data on a per pathogen basis in S1 Table. Overall, the studies evaluated 26 different pathogens. Looking only at experiments that directly compared SBS to EHM, SBS failed to detect pathogens in 29% cases where EHM detected a pathogen, while EHM failed only in 3% of cases where SBS detected a pathogen (Fig 4). This comparison includes 12 different publications, where there were 20 instances where SBS failed to detect 11 different pathogen sof interest and EHM succeeded (Fig 4 and Table 2).
This figure summarizes detailed data found in S1 Table. It examines only head-to-head experiments which directly compared soiled bedding sentinels (SBS) to environmental health monitoring (EHM). a) Confusion matrix summarizing these experiments, given the differing methodologies involved. Thus, when EHM detected a pathogen, SBS failed in 29% of these cases (i.e., SBS showed a sensitivity of 71% relative to EHM), see EHM “Yes” row. Conversely, when SBS detected a pathogen, EHM failed in 3% of cases (i.e., EHM showed a sensitivity of 97% relative to SBS), see SBS “Yes” column. b) Breakdown, by the pathogens evaluated, where experiments found differing results between EHM and SBS (pathogens with perfect agreement are excluded). The number of these experiments where SBS detected the pathogen / the number of where EHM detected the pathogen is shown in order of relative SBS performance.
Across all studies, we considered a health monitoring technique to be effective if it detected the pathogen at least some of the time by at least two articles, consistent with the criteria established by de Bruin et al. Based on this standard, EDT and SFSB appear to be effective at detecting 10 pathogens including mouse norovirus (MNV), mouse parvovirus (MPV), Helicobacter spp., Proteus mirabilis, Rodentibacter spp., Staphylococcus spp., Ectoparasites, Entamoeba spp., Spironucleus spp., and pinworms (Aspiculuris spp. & Sypacia obvelata). Additionally, EDT appears to be effective for 6 additional pathogens including mouse hepatitis virus (MHV), murine astrovirus, Tritrichomonas spp., Klebsiella spp., Corynebacterium bovis, and Pneumocystis spp, and Tritichomonas spp. Finally, several other pathogens were only evaluated in a single paper as shown in Table 2 and therefore no broad conclusions were drawn.
Study characteristics
The detailed characteristics of the 42 studies included are present in S2 Table with a high-level summary provided in Table 3. The design of studies varied substantially including 38 experimental studies and 5 case studies. It included publications investigating EDT (n = 27), SFSB (n = 8), and DCS (n = 24). Studies were split as to whether they investigated multiple EHM methods (n = 17), a single EHM method (n = 25), and whether they compared at least one EHM method to SBS (n = 22). As mentioned above, the statistical methods used are robust with respect to missing data, are well suited to this heterogeneity, and were extensively stress-tested regardless. The studies included three articles that did not have data suitable for quantitative statistical analysis (one article was a cost analysis, one was a protocol, and one was the sole article detailing REM).
Of the articles using EDT and SFSB, a variety of materials and methods were used to collect samples. For EDT, 14 studies used more than one sampling material type. These sampling materials included media (n = 22), non-adhesive swabs (n = 11), and adhesive swabs (n = 5). For SFSB, 5 studies used more than one sampling material type. These sampling materials included media (n = 6), non-adhesive swabs (n = 5), and adhesive swabs (n = 2). SFSB exposure methods included agitation (n = 5), dredging (n = 3), or stagnant exposure (n = 3).
All studies were performed with mice. For mouse characteristics, 18 studies used only female mice, 9 used both sexes, 3 used only males (11 did not report mouse sex and 1 was a cost analysis.) Additionally, 18 studies used a common outbred stock (Swiss Webster, CD1, or ICR), 5 used C57BL/6, 3 used nude (Foxn1nu/Foxn1nu) ta mice, and 16 used other mouse strains or stocks (various vendors). One study did not report mouse strain. Mouse cages contained bedding made of corncob (n = 17), wood (n = 17), paper (n = 2), or wood/paper (n = 1), when reported. All but four articles used IVCs to house the animals (those used static caging). Of the IVCs, most used rack-level filtration (n = 30) and some cage-level filtration (n = 8). The number of cages used within each article ranged from 2 to 21,000, with an average of 1,814 cages (SD = 5766).
Cage or rack ventilation and sanitation also varied greatly and were not always reported. Most often, 60 to 70 IVC air changes per hour were reported (n = 15), 30 to 45 air changes per hour were also sometimes used (n = 4), and 8 reported other values. Rack sanitation, when reported, was often achieved using a rack washer (n = 14) and/or autoclaving (n = 14). Cage sanitation, when reported, was often achieved with a mechanical washer (n = 9) and/or autoclaving (n = 19).
Terminology
The included articles used a wide variety of terms to describe rodent health monitoring and environmental health monitoring as distinct concepts. All 42 articles used the term “monitoring,” 38 articles used the word “health,” and 34 used the compound term “health monitoring.” Additional common terms included “microbiological” or “microbiota” monitoring (n = 19), “hygiene” or “hygienic” (n = 12), “surveillance” (n = 14), or “routine” (n = 14). For environmental health monitoring, 30 articles used the term “environmental” at least once, while the other 13 only addressed monitoring strategies explicitly (i.e., “exhaust air monitoring”). Of the 30 articles that mentioned exhaust dust testing, 27 used the term “exhaust,” 26 used the term “air,” 20 used “dust,” and 16 used “monitoring.”
Bias assessment
In the risk of bias assessment (results found in S2 Table), 12 articles were excluded due to only evaluating a single method for rodent health monitoring, being a case report, or a cost analysis. The following statements refer to only articles included in the bias assessment. Most articles had similar groups at baseline (n = 27), and appeared to be free of other problems that could result in bias (n = 28). Additionally, for most articles it was neither applicable for animals/cages to be selected at random (n = 28) nor did they have incomplete outcome data that needed to be addressed (n = 25). For most articles it was unclear if outcome assessors were blinded (n = 26). All articles appeared to be free of selective outcome reporting.
Discussion
Environmental health monitoring has superior pathogen detection over soiled bedding sentinels
Overall, the data from this systematic review shows that EHM has superior pathogen detection over that of SBS. This conclusion is supported by both logistic regression and descriptive statistics regardless of pathogen taxonomic classification (such as between viruses and ectoparasites), pathogen importance (such as between agents recommended to be tested every 3-months/annually versus optionally), or type of EHM (such as between EDT, SFSB, or DCM). Therefore, EHM has improved detection over SBS for the majority of agents on each institution’s exclusion list and this holds true regardless of the classification and testing frequency. Across all of the articles identified and included in the systematic review, at least two peer-reviewed articles have shown that at least one type of EHM method can detect 18 different pathogens across taxonomic classifications. In particular, there is good evidence for EDT consistently detecting 16 pathogens and SFSB detecting 10 pathogens, again across taxonomic classification. When compared to the only systematic literature of SBS methods, SBS only demonstrated good evidence of detecting 5 pathogens [1].
The studies reviewed include investigations in different facilities, continents, bedding types, mouse strains, caging/rack types, and ventilation changes. Despite this diversity in their exact components and study designs, generally consistent conclusions were met in that EHM is superior to SBS. Still, we recommend that future research should accurately report all potentially relevant housing factors for reference.
Overall, there were 21 cases where SBS failed to detect a particular pathogen compared to 6, 1, and 3 cases where EDT, SFSB, and DCS failed to detect a particular pathogen, respectively. To follow, we will discuss the publications in which EHM methods failed to detect particular pathogens, as these studies are sometimes cited as rationale against switching to EHM. Some of these articles had higher risks of bias as indicated in S2 Table.
In Compton 2004, EDT failed to detect Murine Rotavirus [12]; however, in this case, soiled bedding sentinels also failed to detect this pathogen and therefore this finding cannot be used to demonstrate the superiority of SBS.
In Mahabir 2019, EDT failed to detect Klebsiella spp., Staphylococcus spp., and Streptococcus spp. [6]; however, in the discussion authors note that this was likely due to low prevalence and/or organism numbers of Staphylococcus spp. (5–9 cages) and Klebsiella spp. (1 cage), as well as the absence of shedding of Streptococcus spp. In fact, they note that at the end of the study Streptococcus spp. was not detected in any mouse. Additionally, SBS was not evaluated so it’s unknown if it would have detected these pathogens. We note that EDT has successfully been found to detect Staphylococcus spp. in 6 additional papers [44, 52, 53, 55, 57, 61] and Klebsiella spp. in 2 publications [52, 55].
Bauer 2016 is often cited as demonstrating that EHM is not effective as the paper concluded that EDT consistently failed to detect MNV (and only inconsistently detected MPV) [33]. At the time of publication, the authors did not have an explanation for this finding. However, the literature and methods behind EDT have evolved since then. A likely explanation is that early testing methods used in this paper were less sophisticated than current standards. In particular, the type of filter paper used was rudimentary and the size of the filter paper was smaller than modern materials which may have led to lack of detection. Additionally, this study did have SBS as a control condition and it is possible the infected mice were no longer shedding in the case of MPV. We note that MNV has successfully been detected by EDT in 5 additional publications [7, 10, 50, 52, 61].
In Miller 2018, EDT failed to detect Pseudomonas aeruginosa when SBS did detect this pathogen [52]. In this case, authors hypothesized this could be due to pooling, infected mice being removed before gauze pieces were installed (gauze were only present for 3 weeks), or a false positive in bacterial culture or a transient infection that did not lead to shedding. Pseudomonas aeruginosa was detected by EDT in Ragland 2019 [55].
Finally, in Winn 2022, SFSB and DCS failed to detect beta-haemolytic Streptococcus while SBS did detect it [62]; this is likely due to the prevalence of this opportunistic agent being below the sensitivity for the assay, which was indicated by low copy numbers attained via SBS. Additionally, in Winn 2022, DCS failed to detect Klebsiella spp. or Proteus mirabilis, again, likely due to lower organism copy numbers in that study; however, we note that SFSB was able to detect these pathogens, and DCS is generally recommended as a supplementary rodent health monitoring strategy.
A systematic review is only as good as the studies that are included in it, and due to the differences in testing parameters for animal versus replacement studies, the formal risk of bias assessment we performed was limited, although indicated relatively low risk of bias across studies. Furthermore, as the studies reviewed in this manuscript do not deal with determination of causality (and instead are focused on the ability of a technique to detect), there is much less risk of bias. There are no animals to assign to treatment groups or caretakers to blind to treatment allocation. The biggest risk for bias comes from publication bias (i.e., not publishing negative results). Given that the majority of the studies compared SBS to EHM, this is unlikely as both methods would have to fail to detect all pathogens, whereas a failure of EHM to detect a pathogen when SBS was successful would be a more alluring finding for publication. In the smaller group of papers that only examined EHM, the risk of publication bias is arguably higher. However, including or excluding these papers do not affect the outcome of our statistical analyses, arguing against any significant bias.
Ethical, scientific, and practical implications
Rodent health monitoring programs are intended to assist research laboratories in detecting and/or determining absence of adventitious pathogens, but the reliability and success rate of methods used is now receiving more scrutiny. Despite SBS programs having been the traditional practice for many years, the 2016 systematic review on the efficacy of this methodology only found evidence of ability to detect 5 pathogens [1].
Conversely, the results of this review clearly demonstrate the superiority of EHM methods in pathogen detection. Therefore, we argue that there is a scientific and ethical imperative for programs to work towards replacing SBS with EHM. Full implementation of the 3Rs (Refinement, Reduction, and Replacement) is considered an essential part of conducting humane science. There were 4 papers included in this review that mentioned the specific number of animals that would be replaced at their facilities due to switching to EHM [7, 10, 23, 52]. Across just these four articles a total of 6,876 animals will be replaced yearly. In our recent survey, 111 institutions indicated the potential for a total of well over 20,000 additional rodents to be replaced each year [23].
Beyond the positive ethical and scientific implications of EHM, there was often frequent discussion of cost and time savings to the technique. Thirteen articles mentioned cost savings and this review includes one formal cost analysis [23]. An additional article indicated the technique reduced costs 3-fold [7]. Additionally, six articles mentioned time savings, indicating benefits from not needing to check sentinel mice, spending less time on cage changes/care and diagnostic testing procedures.
Of course, dismantling and replacing traditional health monitoring programs takes time and effort. Our recent cross-sectional survey indicates that although personnel generally believe EHM is good, they do not feel professional pressure to make the switch to EHM and are not confident in their ability to make the switch [22]. Additionally, all three factors are positively associated with intentions to switch to EHM. It is therefore imperative to create pressure to switch to EHM and provide resources to increase confidence in making the switch. To support this aim, the 3Rs Collaborative has created an extensive resource hub on the topic that includes basic information about EHM, lists names and logos of institutions that have switched, editable standard operating procedures, an editable slide deck for internal presentations to gain consensus for the switch, sanitation information, cost analysis information, frequently asked questions, information on how to switch, and records of prior presentations (www.3rc.org/ehm).
Although we recommend EHM as the core part of rodent health monitoring, it should not be considered a substitution for techniques required to detect and characterize novel agents. After all, as EHM is reliant on PCR it can miss any agents that are novel or not included in the testing panel. Therefore, if abnormal clinical signs or study results are observed in colony rodents without a corresponding diagnosis via the EHM panel, we recommend performing a thorough diagnostic workup (e.g., histopathology, clinical chemistry, blood count) on the affected colony animals.
Concordance is needed on terminology for environmental health monitoring
Our findings in this article and general expertise in the field highlight the need for concordance on terminology for EHM. The success of EHM calls for a paradigm shift in the methods used for routine rodent health monitoring. To assist in this shift to EHM, clear, and accurate communication is needed. While conducting this systematic review, it was clear that there was a need for both new standardized terminology and an organizational structure for what constitutes EHM. At its core, and by our definition, EHM involves collecting a sample for routine health monitoring without the need for live sentinel animals.
After reviewing the literature and having consensus discussions on the frequency of terminology in articles, vendor-specific terminology, and technique-specific factors our expert working group created a table of preferred terminology (Table 1). Thus far, we divide EHM into 4 main types based on sample collection. These include exhaust dust testing (EDT) within IVC racks, sampling pooled soiled bedding by a variety of techniques (SFSB), sampling dust from the housing room or husbandry equipment (REM), and finally the sampling of the intra-cage environment of individual cages, or minimally invasive sampling of individual research animals by direct colony sampling (DCS). For EDT and SFSB, various variations to the primary method are being investigated to increase detection sensitivity while still being practical. As EHM techniques are refined over time, we strongly encourage laboratory animal veterinarians and managers to consistently use these vendor-agnostic terms to advance knowledge more quickly and allow research to be identified quickly by other professionals.
Future research–targets and best practices
As of October of 2023 (the end of the sampling period for this article), only 8 publications had reported outcomes from SFSB sampling. This is likely due to the lag in the development of this technique compared to EDT with reports of SFSB validation only appearing since 2017. So far, these publications have reported using a variety of sampling materials and methods (although most with success). Still, further research is needed to determine what specific materials and methods are most effective, especially since this method can be used on all caging/rack types.
In particular, there may be concern with respect to monitoring mice in specific opportunistic pathogen free and/or immunodeficient health status settings, especially where infection rates or prevalence may be low. Thus far, there is good support to demonstrate the ability of EDT to detect and help eliminate Corynebacterium bovis from immunodeficient mice within barrier rooms [8, 49] and that SFSB can detect murine astrovirus and other pathogens in specific pathogen free mice [62]. Continued investigations may be beneficial to demonstrate the utility of EHM in areas with extensive exclusion lists.
Independent of the health monitoring methods used, it would also be beneficial to conduct research on the effect of sample pooling on pathogen detection by PCR assays. Pooling refers to operating procedures where individual samples collected from a cohort (generally fecal pellets, oral swabs, or fur swabs) are combined and processed as one sample. For EDT and SFSB, pooling typically refers to combining media and/or bedding from multiple racks. Pooling presents several practical challenges. For instance, the volume of samples that can be combined into one test is physically limited; and/or sample homogeneity can suffer at greater pooling volumes [64]. Additionally, there is the possibility that pooled samples dilute the signal for low-prevalence infections. Further research in pooling samples by DCS as a direct comparison to or supplement with other environmental health monitoring methods is needed. Regardless, all such methods can continue to replace the use of sentinel animals.
For any future research on EHM, we recommend the following best practices. First, a gold standard control—such as a naïve contact sentinel—should be used within each study to demonstrate if the pathogen of interest is being shed and, more importantly, indicate whether the pathogen of interest is infectious and leading to infection at the peak of the shedding period [11, 18, 65, 66]. This is necessary as it is possible for infected mice to shed neutralized virus and direct testing of infected mice will indicate the presence of a virus even though it is no longer of a concern to the broader colony. Second, future studies should follow both the PREPARE guidelines [67] and ARRIVE guidelines [68] in reporting to help ensure good study quality and assist future systematic reviews. These guidelines emphasize essential experimental design concepts like randomization and blinding. For EHM studies in particular, authors should report relevant items such as type of bedding, air changes per hour, sanitation protocols, and type of caging/rack. Third, authors should use the recommended terminology used in this article to assist in future identification of articles.
Limitations
One potential criticism of this systematic review is that we did not include conference posters or presentations about EHM. However, this is best practice for two reasons. First, we avoid the duplication of the same experiment in the data set (due to multiple abstracts at different conferences and/or final peer-reviewed publication). Second, abstracts that do not convert to a peer-reviewed publication can reflect weaker studies that authors feel are useful pilot data but not fit for publication or studies with methodological issues that are detected at peer review.
The wide variety of methodologies used in the papers included in this review precluded completion of a meta-analysis. After all, there are no agreed upon standards for performing health monitoring with either SBS or EHM. For example, SBS programs can vary in sentinel immune status (immunocompetent or compromised), numbers, timing, frequency of testing, type of testing (e.g., serology, PCR, histopathology, bacterial culture, parasitology), sample type, and ratio of colony cages to sentinel cages [69]. Similarly, EHM programs vary in methodology as indicated in Table 1 and in aspects related to frequency of testing, sample type, and ratio of colony cages to media.
Accordingly, it was necessary to turn more discrete outcome variables into yes/no detection, which could have elevated estimates of sensitivity. However, this would have affected SBS and EHM equally, still allowing a fair comparison. Furthermore, this also reflects the reality of health monitoring–a single positive result is typically interpreted as indicative of a broader pathogen presence. Despite these limitations, it is still clear that using EHM is superior to SBS for routine rodent health monitoring.
Conclusion
Our findings highlight the strong evidence-base supporting the superiority of EHM over SBS regardless of pathogen type, pathogen importance level, or specific type of EHM. Of note, even as of 2023 there is a larger evidence base for both EDT and SFSB than there was for SBS in 2016 [1]. This clear superiority in testing efficacy, combined with the 3Rs benefit of Replacement of animals, argues for a scientific and ethical priority to replace SBS with EHM. Cost and labor savings only further the case for EHM.
Supporting information
S1 File. PRISMA checklist.
The 2020 PRISMA checklist with section references to where all relevant information can be found.
https://doi.org/10.1371/journal.pone.0311840.s001
(DOCX)
S2 File. A priori protocol.
The study protocol written out a priori before beginning data collection.
https://doi.org/10.1371/journal.pone.0311840.s002
(PDF)
S3 File. Search strategy.
The full text search strategy used for each database.
https://doi.org/10.1371/journal.pone.0311840.s003
(DOCX)
S1 Table. Pathogen evaluation and recode list.
A list of all pathogens evaluated in each article with conclusions as to whether they were detected and information on how they were coded for analysis.
https://doi.org/10.1371/journal.pone.0311840.s004
(XLSX)
S2 Table. Environmental health monitoring articles.
A detailed table of all articles included in the systematic literature review.
https://doi.org/10.1371/journal.pone.0311840.s005
(XLSX)
S3 Table. Study list.
A complete list of all articles screened in this systematic review, decision as to whether to include, and reason for decision.
https://doi.org/10.1371/journal.pone.0311840.s006
(XLSX)
S1 Dataset. Final dataset and analyses.
This file includes the SAS code for the dataset used in analysis and code for both final analyses as well as notes on the steps taken for each analysis.
https://doi.org/10.1371/journal.pone.0311840.s007
(DOCX)
S2 Dataset. Matched dataset and analyses.
This file includes the SAS code for the matched dataset only including the publications that directly compared EHM and SBS and the corresponding analyses and steps.
https://doi.org/10.1371/journal.pone.0311840.s008
(DOCX)
Acknowledgments
We would like to gratefully acknowledge Lauren Young for assistance with study selection and Jane Yacitilla for early consultation on our search strategy. We also thank the 3Rs Collaborative staff, volunteers, and members for making this research possible. We appreciate all the laboratory animal personnel who have been integral in helping their institutions replace soiled bedding sentinels with environmental health monitoring. Finally, we thank all of the sentinel rodents that have been, and are currently, used for rodent health monitoring programs. Their contribution has helped ensure the validity and reproducibility of scientific findings for several decades. We hope this review helps accelerate their Replacement where appropriate.
References
- 1. Bruin WCC de, Ven EME van de, Hooijmans CR. Efficacy of Soiled Bedding Transfer for Transmission of Mouse and Rat Infections to Sentinels: A Systematic Review. PLOS ONE. 2016;11: e0158410. pmid:27518654
- 2.
Macleod MR, Tanriver-Ayder E, Hair K, Sena E. Design of Meta-Analysis Studies. In: Bespalov A, Michel MC, Steckler T, editors. Good Research Practice in Non-Clinical Pharmacology and Biomedicine. Cham: Springer International Publishing; 2020. pp. 299–317. https://doi.org/10.1007/164_2019_289 pmid:31620915
- 3. Jensen ES, Allen KP, Henderson KS, Szabo A, Thulin JD. PCR Testing of a Ventilated Caging System to Detect Murine Fur Mites. J Am Assoc Lab Anim Sci. 2013;52: 28–33. pmid:23562030
- 4. Dubelko AR, Zuwannin M, McIntee SC, Livingston RS, Foley PL. PCR Testing of Filter Material from IVC Lids for Microbial Monitoring of Mouse Colonies. J Am Assoc Lab Anim Sci. 2018;57: 477–482. pmid:30092857
- 5. Hanson WH, Taylor K, Taylor DK. PCR Testing of Media Placed in Soiled Bedding as a Method for Mouse Colony Health Surveillance. J Am Assoc Lab Anim Sci. 2021;60: 306–310. pmid:33952384
- 6. Mahabir E, Durand S, Henderson KS, Hardy P. Comparison of two prevalent individually ventilated caging systems for detection of murine infectious agents via exhaust air particles. Lab Anim. 2019;53: 84–88. pmid:30012043
- 7. Mailhiot D, Ostdiek AM, Luchins KR, Bowers CJ, Theriault BR, Langan GP. Comparing Mouse Health Monitoring Between Soiled-bedding Sentinel and Exhaust Air Dust Surveillance Programs. J Am Assoc Lab Anim Sci. 2020;59: 58–66. pmid:31862019
- 8. Manuel CA, Pugazhenthi U, Leszczynski JK. Surveillance of a Ventilated Rack System for Corynebacterium bovis by Sampling Exhaust-Air Manifolds. Journal of the American Association for Laboratory Animal Science. 2016;55: 58–65. pmid:26817981
- 9. O’Connell KA, Tigyi GJ, Livingston RS, Johnson DL, Hamilton DJ. Evaluation of In-cage Filter Paper as a Replacement for Sentinel Mice in the Detection of Murine Pathogens. J Am Assoc Lab Anim Sci. 2021;60: 160–167. pmid:33629939
- 10. Pettan-Brewer C, Trost RJ, Maggio-Price L, Seamons A, Dowling SC. Adoption of Exhaust Air Dust Testing in SPF Rodent Facilities. Journal of the American Association for Laboratory Animal Science. 2020;59: 156–162. pmid:32075699
- 11. Lindstrom KE, Carbone LG, Kellar DE, Mayorga MS, Wilkerson JD. Soiled bedding sentinels for the detection of fur mites in mice. J Am Assoc Lab Anim Sci. 2011;50: 54–60. pmid:21333164
- 12. Compton SR, Homberger FR, Paturzo FX, Clark JM. Efficacy of three microbiological monitoring methods in a ventilated cage rack. Comp Med. 2004;54: 382–392. pmid:15357318
- 13. Scharmann W, Heller A. Survival and transmissibility of Pasteurella pneumotropica. Lab Anim. 2001;35: 163–166. pmid:11315166
- 14. Albers TM, Henderson KS, Mulder GB, Shek WR. Pathogen Prevalence Estimates and Diagnostic Methodology Trends in Laboratory Mice and Rats from 2003 to 2020. J Am Assoc Lab Anim Sci. 2023;62: 229–242. pmid:37127407
- 15. Grove KA, Smith PC, Booth CJ, Compton SR. Age-associated variability in susceptibility of Swiss Webster mice to MPV and other excluded murine pathogens. J Am Assoc Lab Anim Sci. 2012;51: 789–796. pmid:23294885
- 16. Besselsen DG, Wagner AM, Loganbill JK. Effect of mouse strain and age on detection of mouse parvovirus 1 by use of serologic testing and polymerase chain reaction analysis. Comp Med. 2000;50: 498–502. pmid:11099132
- 17. Besselsen DG, Myers EL, Franklin CL, Korte SW, Wagner AM, Henderson KS, et al. Transmission probabilities of mouse parvovirus 1 to sentinel mice chronically exposed to serial dilutions of contaminated bedding. Comp Med. 2008;58: 140–144. pmid:18524171
- 18. Rice KA, Albacarys LK, Metcalf Pate KA, Perkins C, Henderson KS, Watson J. Evaluation of diagnostic methods for Myocoptes musculinus according to age and treatment status of mice (Mus musculus). J Am Assoc Lab Anim Sci. 2013;52: 773–781. pmid:24351766
- 19. Barthold SW, Beck DS, Smith AL. Enterotropic coronavirus (mouse hepatitis virus) in mice: influence of host age and strain on infection and disease. Lab Anim Sci. 1993;43: 276–284. pmid:8231082
- 20. Eguíluz C, Rossi M, Viguera E. Pinworm detection in mice with immunodeficient (NOD SCID) and immunocompetent (CD-1 and Swiss) soiled bedding sentinels in individually ventilated cage systems. Lab Anim. 2015;49: 302–310. pmid:25667226
- 21. Kain ML, Arbona RRJ, Henderson KS, Dhawan R, Monette S, Lipman NS. Infectivity and Shedding of Mouse Kidney Parvovirus After Oronasal Inoculation of C57BL/6, CD1, and NSG Mice. Comp Med. 2022;72: 376–385. pmid:36744512
- 22. Luchins KR, Gates KV, Winn CB, Manuel CA, Pettan-Brewer C, Foley PL, et al. A Cross-sectional Survey on Rodent Environmental Health Monitoring Practices: Benchmarking, Associations, and Barriers. J Am Assoc Lab Anim Sci. 2023;62: 64–73. pmid:36755202
- 23. Luchins KR, Bowers CJ, Mailhiot D, Theriault BR, Langan GP. Cost Comparison of Rodent Soiled Bedding Sentinel and Exhaust Air Dust Health-Monitoring Programs. J Am Assoc Lab Anim Sci. 2020;59: 508–511. pmid:32591028
- 24. Marx JO, Gaertner DJ, Smith AL. Results of Survey Regarding Prevalence of Adventitial Infections in Mice and Rats at Biomedical Research Facilities. J Am Assoc Lab Anim Sci. 2017;56: 527–533. pmid:28903823
- 25. Mähler M, Berard M, Feinstein R, Gallagher A, Illgen-Wilcke B, Pritchett-Corning K, et al. FELASA recommendations for the health monitoring of mouse, rat, hamster, guinea pig and rabbit colonies in breeding and experimental units. Lab Anim. 2014;48: 178–192. pmid:24496575
- 26. Bannach-Brown A, Przybyła P, Thomas J, Rice ASC, Ananiadou S, Liao J, et al. Machine learning algorithms for systematic review: reducing workload in a preclinical review of animal studies and reducing human screening error. Systematic Reviews. 2019;8: 23. pmid:30646959
- 27. Eichner M, Smith JM. Assessing Methods for Replacement of Soiled Bedding Sentinels in Cage-level Exhaust IVC Racks. Journal of the American Association for Laboratory Animal Science. 2023;62: 409–415. pmid:37758466
- 28. Manuel C, Luchins K, Peterson C, Dodelet-Devillers A, Pettan-Brewer C, Phaneuf L, et al. Letter to the editor regarding ‘Assessing Methods for Replacement of Soiled Bedding Sentinels in Cage-level Exhaust IVC Racks’ by Eichner M et al. 2023. JAALAS. Under Review.
- 29.
Allison PD. Logistic Regression Using the SAS System: Theory and Application | Wiley. 1st ed. WA (Wiley-SAS); 2001. Available: https://www.wiley.com/en-us/Logistic+Regression+Using+the+SAS+System%3A+Theory+and+Application-p-9780471221753
- 30. Thabane L, Mbuagbaw L, Zhang S, Samaan Z, Marcucci M, Ye C, et al. A tutorial on sensitivity analyses in clinical trials: the what, why, when and how. BMC Medical Research Methodology. 2013;13: 92. pmid:23855337
- 31.
Grafen A, Hails R. Modern Statistics for the Life Sciences. Oxford, New York: Oxford University Press; 2002.
- 32. Ziegler A, Grömping U. The Generalised Estimating Equations: A Comparison of Procedures Available in Commercial Statistical Software Packages. Biometrical Journal. 1998;40: 245–260.
- 33. Bauer BA, Besch-Williford C, Livingston RS, Crim MJ, Riley LK, Myles MH. Influence of Rack Design and Disease Prevalence on Detection of Rodent Pathogens in Exhaust Debris Samples from Individually Ventilated Caging Systems. J Am Assoc Lab Anim Sci. 2016;55: 782–788. pmid:27931317
- 34. Buchheister S, Roegener F, Zschemisch N-H, Talbot SR, Christensen H, Bleich A. One for two: A novel and highly sensitive virulence factor-based quantitative polymerase chain reaction assay for the simultaneous detection of Rodentibacter pneumotropicus and Rodentibacter heylii in environmental sample material. Lab Anim. 2020;54: 239–250. pmid:31195883
- 35. Butt J, Schmitz M, Berkus B, Schmidt K, Höfler D. Validation of Multiplex PCR and Serology Detecting Helicobacter Species in Mice. Microorganisms. 2023;11: 249. pmid:36838214
- 36. Chou C-L, Cheng Y-C, Wang M-H, Wan C-H. Novel pcr assays for three fur mites in naturally infected mice without animal sacrifice. Taiwan Vet J. 2020;46: 133–142.
- 37. Clancy BM, Theriault BR, Schoenberger JM, Bowers CJ, Mitchell CM, Langan GP, et al. Identification and Control of an Ornithonyssus Bacoti Infestation in a Rodent Vivarium by Using Molecular Diagnostic Techniques. Comparative Medicine. 2022;72: 113–121. pmid:35379379
- 38. Compton SR, Macy JD. Effect of Cage-Wash Temperature on the Removal of Infectious Agents from Caging and the Detection of Infectious Agents on the Filters of Animal Bedding-Disposal Cabinets by PCR Analysis. J Am Assoc Lab Anim Sci. 2015;54: 745–755. pmid:26632784
- 39. Compton SR, Booth CJ, Macy JD. Murine Astrovirus Infection and Transmission in Neonatal CD1 Mice. J Am Assoc Lab Anim Sci. 2017;56: 402–411. pmid:28724490
- 40. Gerwin PM, Arbona RJR, Riedel ER, Henderson KS, Lipman NS. PCR Testing of IVC Filter Tops as a Method for Detecting Murine Pinworms and Fur Mites. J Am Assoc Lab Anim Sci. 2017;56: 752–761. pmid:29256370
- 41. Iwantschenko A-K, Roegener F, Garrels W, Dorsch M, Köhl W, Riehle C, et al. Why serology just is not enough: Strategic parvovirus risk assessment using a novel qPCR assay. Lab Anim. 2022;56: 380–395. pmid:35102773
- 42. Jacobsen K, Mahabir E, Brielmeier M, Wilhelm P, Seidel KE, Schmidt J. Monitoring a mouse colony for Helicobacter bilis using a Helicobacter-genus-specific nested PCR. Lab Anim. 2005;39: 400–412. pmid:16197707
- 43. Kapoor P, Hayes YO, Jarrell LT, Bellinger DA, Thomas RD, Lawson GW, et al. Evaluation of Anthelmintic Resistance and Exhaust Air Dust PCR as a Diagnostic Tool in Mice Enzootically Infected with Aspiculuris tetraptera. J Am Assoc Lab Anim Sci. 2017;56: 273–289. pmid:28535863
- 44. Kim H, Jeon C-M, Jang Y-C, Goo J-S, Park J-H. Evaluation of exhaust air dust polymerase chain reaction as a supplement method for soiled bedding sentinel monitoring in specific pathogen free mouse facility using two different individually ventilated cage racks. Lab Anim. 2023;57: 40–49. pmid:36204980
- 45. Körner C, Miller M, Brielmeier M. Detection of Murine Astrovirus and Myocoptes musculinus in individually ventilated caging systems: Investigations to expose suitable detection methods for routine hygienic monitoring. PLOS ONE. 2019;14: e0221118. pmid:31408494
- 46. Leblanc M, Berry K, Graciano S, Becker B, Reuter JD. False-Positive Results after Environmental Pinworm PCR Testing due to Rhabditid Nematodes in Corncob Bedding. Journal of the American Association for Laboratory Animal Science. 2014;53: 717–724. pmid:25650980
- 47. Luchins KR, Mailhiot D, Theriault BR, Langan GP. Detection of Lactate Dehydrogenase Elevating Virus in a Mouse Vivarium Using an Exhaust Air Dust Health Monitoring Program. J Am Assoc Lab Anim Sci. 2020;59: 328–333. pmid:32079556
- 48. Lupini L, Bassi C, Guerriero P, Raspa M, Scavizzi F, Sabbioni S. Microbiota and environmental health monitoring of mouse colonies by metagenomic shotgun sequencing. World J Microbiol Biotechnol. 2022;39: 37. pmid:36472670
- 49. Manuel CA, Pugazhenthi U, Spiegel SP, Leszczynski JK. Detection and Elimination of Corynebacterium bovis from Barrier Rooms by Using an Environmental Sampling Surveillance Program. J Am Assoc Lab Anim Sci. 2017;56: 202–209. pmid:28315652
- 50. Miller M, Ritter B, Zorn J, Brielmeier M. Exhaust Air Particle PCR Detects Helicobacter hepaticus Infections at Low Prevalence. Journal of Veterinary Science & Technology. 2016;7.
- 51. Miller M, Ritter B, Zorn J, Brielmeier M. Exhaust Air Dust Monitoring is Superior to Soiled Bedding Sentinels for the Detection of Pasteurella pneumotropica in Individually Ventilated Cage Systems. J Am Assoc Lab Anim Sci. 2016;55: 775–781. pmid:27931316
- 52. Miller M, Brielmeier M. Environmental samples make soiled bedding sentinels dispensable for hygienic monitoring of IVC-reared mouse colonies. Lab Anim. 2018;52: 233–239. pmid:29145766
- 53. Niimi K, Maruyama S, Sako N, Miyata K, Yoshimoto T, Bilecki B, et al. The SentinelTM EADR program can detect more microorganisms than bedding sentinel animals. JJVR. 2018;66: 125–129.
- 54. Parkinson CM, O’Brien A, Albers TM, Simon MA, Clifford CB, Pritchett-Corning KR. Diagnosis of Ecto- and Endoparasites in Laboratory Rats and Mice. JoVE (Journal of Visualized Experiments). 2011; e2767. pmid:21912374
- 55. Ragland NH, Miedel EL, Engelman RW. PCR Prevalence of Murine Opportunistic Microbes and their Mitigation by Using Vaporized Hydrogen Peroxide. J Am Assoc Lab Anim Sci. 2019;58: 208–215. pmid:30795821
- 56. Scavizzi F, Bassi C, Lupini L, Guerriero P, Raspa M, Sabbioni S. A comprehensive approach for microbiota and health monitoring in mouse colonies using metagenomic shotgun sequencing. Animal Microbiome. 2021;3: 53. pmid:34325744
- 57. Schlapp G, Fernández-Graña G, Arévalo AP, Crispo M. Establishment of an environmental microbiological monitoring program in a mice barrier facility. An Acad Bras Ciênc. 2018;90: 3155–3164. pmid:30304242
- 58. Srinivasan MR, Vijay K, Karuppannan AK, Ramesh S, Krishnamohan RY. Non-Invasive Fecal Based PCR Assays for Detection of Mouse Parvo Virus and Minute Virus of Mice in Laboratory Mice. Indian Journal of Animal Research. 2021;55: 1356–1359.
- 59. Srinivasan MR, Vijay K, Karuppannan AK, Ramesh S, Reddy YK. RT-PCR Assay for Detection of Enterotropic Strains of Mouse Hepatitis Virus in Faecal Samples. Indian Journal of Animal Research. 2022;56: 633–636.
- 60. Tierce RK, Winn AA, Albers TM, Poueymirou WT, Levee EM, Woods SE, et al. Detection and Transmission of Proteus mirabilis in Immunodeficient Mice. Journal of the American Association for Laboratory Animal Science. 2022;61: 256–269. pmid:35277210
- 61. Varela MMD, Bibay JIA, Ogden BE, Crim MJ, Htoon HM. Using Sterile Flocked Swabs as an Alternative Method for Rodent Health Monitoring. Journal of the American Association for Laboratory Animal Science. 2022;61: 370–380. pmid:35764385
- 62. Winn C, Rogers R, Keenan R, Gerwin P, Matthews K, Ramirez J, et al. Using Filter Media and Soiled Bedding in Disposable Individually Ventilated Cages as a Refinement to Specific Pathogen-free Mouse Health Monitoring Programs. JAALAS. 2022;61: 361–369. pmid:35750479
- 63. Zorn J, Ritter B, Miller M, Kraus M, Northrup E, Brielmeier M. Murine norovirus detection in the exhaust air of IVCs is more sensitive than serological analysis of soiled bedding sentinels. Lab Anim. 2017;51: 301–310. pmid:27440411
- 64. Manuel CA, Hsu CC, Riley LK, Livingston RS. Soiled-bedding sentinel detection of murine norovirus 4. J Am Assoc Lab Anim Sci. 2008;47: 31–36. pmid:18459710
- 65. Otto G, Tolwani RJ. Use of microisolator caging in a risk-based mouse import and quarantine program: a retrospective study. Contemp Top Lab Anim Sci. 2002;41: 20–27. pmid:11860254
- 66. Shek WR. Role of housing modalities on management and surveillance strategies for adventitious agents of rodents. ILAR J. 2008;49: 316–325. pmid:18506065
- 67. Smith AJ, Clutton RE, Lilley E, Hansen KEA, Brattelid T. PREPARE: guidelines for planning animal research and testing. Lab Anim. 2018;52: 135–141. pmid:28771074
- 68. Kilkenny C, Browne W, Cuthill IC, Emerson M, Altman DG. Animal research: reporting in vivo experiments: the ARRIVE guidelines. British journal of pharmacology. 2010;160: 1577–1579. pmid:20649561
- 69. Lipman NS, Homberger FR. Rodent Quality Assurance Testing: Use of Sentinel Animal Systems. Lab Anim. 2003;32: 36–43. pmid:19757615