Assessing stakeholders’ opinions on sharing antimicrobial susceptibility testing data from animals is key to developing a centralized database and dashboard tool

Raissa R. Raineri Department of Veterinary Microbiology and Preventive Medicine, College of Veterinary Medicine, Iowa State University, Ames, IA
National Institute of Antimicrobial Resistance Research and Education, Ames, IA

Search for other papers by Raissa R. Raineri in
Current site
Google Scholar
PubMed
Close
 DVM
,
Chelsey B. Shivley USDA APHIS, Portland, ME

Search for other papers by Chelsey B. Shivley in
Current site
Google Scholar
PubMed
Close
 DVM, PhD, DACAW
,
Paul J. Plummer Department of Veterinary Microbiology and Preventive Medicine, College of Veterinary Medicine, Iowa State University, Ames, IA
National Institute of Antimicrobial Resistance Research and Education, Ames, IA
Department of Veterinary Diagnostic and Production Animal Medicine, College of Veterinary Medicine, Iowa State University, Ames, IA

Search for other papers by Paul J. Plummer in
Current site
Google Scholar
PubMed
Close
 DVM, PhD, DACVIM
, and
Amanda J. Kreuder Department of Veterinary Microbiology and Preventive Medicine, College of Veterinary Medicine, Iowa State University, Ames, IA
National Institute of Antimicrobial Resistance Research and Education, Ames, IA

Search for other papers by Amanda J. Kreuder in
Current site
Google Scholar
PubMed
Close
 DVM, PhD, DACVIM

Click on author name to view affiliation information

Abstract

OBJECTIVE

To present, analyze, and discuss stakeholders’ opinions regarding sharing antimicrobial susceptibility testing (AST) data from animals into a centralized database and dashboard tool that would collect, aggregate, store, and analyze this type of data from veterinary diagnostic laboratories (VDLs) across the country.

SAMPLE

1 in-person focus group (9 participants), 9 virtual focus groups (49 participants), and online pre- and postmeeting surveys (76 and 35 participants, respectively).

METHODS

Focus groups and surveys were conducted to assess the opinions of veterinarians, producers, researchers, diagnosticians, and government officials.

RESULTS

A strong majority of stakeholders recognize AMR as a serious concern for both human and animal health and see several benefits in establishing a centralized AMR database; however, several concerns were raised associated with data confidentiality, security, curation, and harmonization. In the interest of alleviating those concerns, among other items, stakeholders suggested education and training of data users, providers, and the public in addition to assuring strong data confidentiality protections.

CLINICAL RELEVANCE

Stakeholder engagement is a critical component of all stages of development, implementation, and utilization of an AMR database and dashboard tool that could be used to inform antimicrobial stewardship in veterinary medicine. This assessment of stakeholders’ needs and concerns can be used to help guide future recommendations for data legal protections as well as a data confidentiality and security framework. Maintaining open communication on data usage, storage, and security as well as involvement and education of data providers, users, and the public will remain key to enabling development of an AMR database and dashboard tool for domesticated animals.

Abstract

OBJECTIVE

To present, analyze, and discuss stakeholders’ opinions regarding sharing antimicrobial susceptibility testing (AST) data from animals into a centralized database and dashboard tool that would collect, aggregate, store, and analyze this type of data from veterinary diagnostic laboratories (VDLs) across the country.

SAMPLE

1 in-person focus group (9 participants), 9 virtual focus groups (49 participants), and online pre- and postmeeting surveys (76 and 35 participants, respectively).

METHODS

Focus groups and surveys were conducted to assess the opinions of veterinarians, producers, researchers, diagnosticians, and government officials.

RESULTS

A strong majority of stakeholders recognize AMR as a serious concern for both human and animal health and see several benefits in establishing a centralized AMR database; however, several concerns were raised associated with data confidentiality, security, curation, and harmonization. In the interest of alleviating those concerns, among other items, stakeholders suggested education and training of data users, providers, and the public in addition to assuring strong data confidentiality protections.

CLINICAL RELEVANCE

Stakeholder engagement is a critical component of all stages of development, implementation, and utilization of an AMR database and dashboard tool that could be used to inform antimicrobial stewardship in veterinary medicine. This assessment of stakeholders’ needs and concerns can be used to help guide future recommendations for data legal protections as well as a data confidentiality and security framework. Maintaining open communication on data usage, storage, and security as well as involvement and education of data providers, users, and the public will remain key to enabling development of an AMR database and dashboard tool for domesticated animals.

Introduction

Acquired antimicrobial resistance (AMR) occurs when medicines are no longer effective in combating infections caused by bacteria, viruses, fungi, or parasites due to genetic changes in those organisms.1 AMR acquisition may happen naturally over time, but some actions like overuse and inappropriate use of antimicrobials contribute to accelerating this process, currently placing AMR in 1 of the top 10 global threats to public health.1,2 Surveillance activities supported by international and US organizations are designed to generate information about AMR. However, those initiatives are generally complex, expensive, and time-consuming to implement; therefore, leveraging and combining existing data can be valuable.3,4 Veterinary diagnostic laboratories (VDLs) across the country routinely perform antimicrobial susceptibility testing (AST) on a wide array of samples, and some perform whole genome sequencing (WGS) of pathogens. Several projects currently aggregate very limited amounts of VDL AST data5,6; however, by leveraging the full scope of AMR data produced by VDLs, a comprehensive database and dashboard tool could be developed to assist the research and application of AMR stewardship efforts in animals.

Stakeholder engagement is a critical component of all stages of development, implementation, and utilization of an AMR dashboard. The understanding of stakeholder needs and concerns, specifically those of data providers, is important to help guide future recommendations for data legal protections as well as a data confidentiality and security framework. Therefore, the purpose of this article is to present, analyze, and discuss stakeholders’ (veterinarians, producers, researchers, diagnosticians, and government officials) opinions regarding sharing AST data from animals into a centralized database and dashboard tool that would collect, aggregate, store, and analyze this type of data from VDLs across the country.

Methods

Iowa State University and the National Institute of Antimicrobial Resistance Research and Education conducted 10 focus groups (1 in person and 9 virtual), as well as online pre- and postmeeting surveys, to capture opinions from different stakeholder groups and characterize their perceived benefits, concerns, and willingness to share AMR-related data into a centralized database. The instruments utilized to collect those inputs were designed to be complementary of each other, and participation was anonymous and voluntary (Supplementary Material S1).

Recruitment and data collection

Institutional review board approval

Iowa State University obtained approval and exempt status from the Institutional Review Board at Iowa State University (IRB# 22-089) for human subjects’ research prior to initiation of this study.

In-person focus group

A single in-person focus group was held at the National Institute for Animal Agriculture (NIAA) annual meeting in Kansas City, Missouri, on April 20, 2022. Recruitment for this focus group was through NIAA conference emails to both members and registered attendees. Hosting this focus group in association with the NIAA meeting was based on a desire to ensure involvement of all facets of animal agriculture, with a special emphasis on recruitment of livestock producers. Attendance was open to anyone who registered for the conference.

At the meeting, paper copies of response forms were distributed to attendees (Supplementary Material S2). Questions were posed verbally by the moderators, attendees were asked to share their initial responses on the paper form, and a group discussion of the responses was then facilitated. The meeting was not recorded and notes on the verbal discussion were collected. The completed forms were gathered at the end of the focus group, and the responses were compiled into a spreadsheet for subsequent data analysis. The focus group instrument initially developed for the in-person focus group was then adapted for the virtual sessions.

Virtual focus groups

A total of 9 virtual focus groups were held via Zoom (Zoom Video Communications Inc) from April 25 to May 24, 2022, and utilized PowerPoint software (Microsoft Corp) as an instrument of presentation (Supplementary Material S3). A moderator script was developed for these meetings to ensure consistency across focus groups and guide the discussion. Potential stakeholders were identified from a wide range of sources, and the final recruitment list totaled over 800 individuals (Supplementary Material S4). An invitation to join one of the 1.5-hour virtual focus groups was then sent via email by National Institute of Antimicrobial Resistance Research and Education leadership to the recruitment list. Individuals interested in the project were able to choose the category with which they best aligned (data providers [veterinarians, producers], VDL diagnosticians, government officials, or researchers) and were asked to participate in 1 of the 3 virtual focus groups held for each of the categories according to their availability.

Each virtual focus group began with an orientation to preserve participants’ anonymity.7 Participants were requested to keep cameras off and provided instructions on how to change their names according to the group they represented (eg, V for veterinarians, P for producers), animal species they worked with (eg, B for beef cattle, H for human), and a random number. During the focus groups, the primary participation was via the Annotate tool within Zoom using either the text or stamp function. In some cases, participants also provided opinions agreeing or disagreeing with other participants’ comments by placing a stamp near it. Screenshots were captured at the end of each activity. Stakeholders were also able to verbally share an opinion and use the chat function if desired. Video and audio (assuring anonymity) of each focus group were recorded for analysis. When clarification was needed during the focus group, occasional nonstandardized questions were asked by moderators.

Pre- and postmeeting surveys

Online surveys were developed using Qualtrics XM software (Qualtrics) and designed to be completed in 15 to 20 minutes to minimize the time commitment of participants. Premeeting surveys were sent to all registered participants of the virtual focus groups and to an email list maintained by NIAA of members and past conference attendees who had indicated interest in AMR. Postmeeting surveys were sent only to registered attendees of the focus groups. While participants were highly encouraged to submit both the pre- and postmeeting surveys, completion was not a requirement to join the focus groups, and response rates were not monitored.

Data analysis

Data generated by open-ended questions together with participants’ comments were classified as qualitative data and analyzed via NVivo software (QSR International). There are 2 main strategies to analyze qualitative data: deductive and inductive approaches.8 We utilized the inductive approach, in which the researcher reads all participants’ statements, identifies common themes, and groups them. For each open-ended question, the process of reading all statements and assigning them to a code was done several times, especially when a new code was created, to ensure that all responses were appropriately assigned. Following inductive coding, tables were used to present the results as percentage of comments related to each theme. Some opinions were linked to more than 1 theme, so some tables do not reflect the actual number of comments but instead the total of comments that could be associated with each theme. In addition, when participants demonstrated agreement with statements made by other participants, the results were tabulated by counting the number of stamps placed by participants beside the corresponding option. Quantitative data analysis of items such as multiple-choice questions that did not require a written comment were evaluated in Excel (Microsoft Corp). Descriptive statistics (frequencies and percentages) were used to summarize data analysis results. In some cases, categories that represented agreement (ie, strongly agree, somewhat agree) or disagreement (ie, strongly disagree, somewhat disagree) were collapsed and reported together; this is documented in the text when applicable.

Results

A summary of the overall approach and the number of participants in each stage of the study is presented (Figure 1).

Figure 1
Figure 1

Overview of project and participation. AMR = Antimicrobial resistance. NAHLN = National Animal Health Laboratory Network. NIAA = National Institute for Animal Agriculture. NIAMRRE = National Institute of Antimicrobial Resistance Research and Education. VDL = Veterinary diagnostic laboratory.

Citation: Journal of the American Veterinary Medical Association 262, 2; 10.2460/javma.23.07.0426

Premeeting surveys

Participant demographics

Premeeting surveys engaged 76 participants, but not all questions were fully completed. Attendees’ demographic data were collected (Supplementary Table S1), and participants were asked to identify their role by selecting all options that applied. They self-identified as veterinarians (40), academics (33), government employees (24), animal health professionals (16), part of the human health sector (2), livestock producers (2), and others (7) and represented all sectors of animal agriculture listed. Ten participants affirmed that they owned livestock, though only 2 identified themselves as livestock producers.

Responses associated with AMR

Level of agreement with belief statements—The level of agreement with several belief statements regarding AMR was assessed, and belief statements were separated into positive and negative statements (Table 1). Results show that there was a strong agreement about “AMR being a serious concern for human health” (85.3%) and “animal agriculture” (66.2%) and demonstrates the acknowledgment among stakeholders that “sharing AMR data into a centralized database is important to combating development of future AMR” (71.6%). Opinions from all 68 participants were divided regarding the possibility of “personal information being compromised by sharing AMR data” (35.3% strongly/somewhat agreed and 41.1% strongly/somewhat disagreed) and about the “chances of negative effects for animal agriculture due to government storing AMR data” (33.8% strongly/somewhat agreed and 45.6% strongly/somewhat disagreed). In addition, most participants did not believe that “government having specific AMR data from a farm is an invasion of privacy” (67.2% strongly/somewhat disagreed). When analyzing government and nongovernment employees separately, only 4.5% of participants who self-identified as government employees strongly/somewhat agreed that this was “an invasion of privacy”; however, this percentage was higher for participants that self-identified as nongovernment (26.7% strongly/somewhat agreed).

Table 1

Level of agreement regarding belief statements associated with AMR from premeeting surveys (n = 68).

I believe that … No. of answers (%)
Strongly agree Somewhat agree Neither agree nor disagree Somewhat disagree Strongly disagree
Positive statements that could indicate support for development of an AMR dashboard
… AMR is a serious concern for human health 58 (85.3%) 5 (7.4%) 3 (4.4%) 1 (1.5%) 1 (1.5%)
… sharing veterinary AMR data from across the country into a centralized database is an important step in combating the development of future AMR1 48 (71.6%) 12 (17.9%) 3 (4.5%) 2 (3.0%) 2 (3.0%)
… AMR is a serious concern for animal agriculture 45 (66.2%) 19 (27.9%) 1 (1.5%) 2 (2.9%) 1 (1.5%)
Negative statements that could indicate concern for development of an AMR dashboard
… centralized collection of AMR data could be used against animal agriculture 20 (29.9%) 23 (34.3%) 9 (13.4%) 11 (16.4%) 4 (6.0%)
… in general1 centralized collection of AMR data could be used against an individual commodity (eg, beef, dairy, poultry, pork, etc) 19 (27.9%) 19 (27.9%) 11 (16.2%) 13 (19.1%) 6 (8.8%)
… AMR data might be released under a FOIA request 17 (25.0%) 21 (30.9%) 24 (35.3%) 1 (1.5%) 5 (7.4%)
… sharing AMR data could lead to data security issues compromising my personal information 8 (11.8%) 16 (23.5%) 16 (23.5%) 19 (27.9%) 9 (13.2%)
… if AMR data is stored within a government agency it is more likely that there will be negative effects for animal agriculture 7 (10.3%) 16 (23.5%) 14 (20.6%) 13 (19.1%) 18 (26.5%)
… the government having specific AMR data from my farm is an invasion of my privacy1 4 (6.0%) 9 (13.4%) 9 (13.4%) 19 (28.4%) 26 (38.8%)

1Questions that have n = 67.

FOIA = Freedom of Information Act.

In contrast, participants were concerned with the idea that “AMR data might be released under a Freedom of Information Act [FOIA] request” (55.9% strongly/somewhat agreed) and either somewhat or strongly agreed with the beliefs that “centralized collection of AMR data could be used against animal agriculture” (64.2%) and “against an individual commodity” (55.8%). Participants who self-identified as government employees were more concerned about “AMR being released under FOIA requests” (65.2% strongly/somewhat agreed) compared to those who did not self-identify as government (51.1% strongly/somewhat agreed). Nongovernment participants primarily selected “neither agree nor disagree” with this statement (42.2% of responses), which suggests that a lack of understanding of the FOIA process may have influenced their response as compared to those of governmental employees, who would have a more complete understanding of the process.

Level of comfort sharing specific AMR data items—Stakeholders were asked to share their level of comfort related to specific data items that might be shared by VDLs (Figure 2). Participants demonstrated extreme comfort sharing information on type of “AST panel test used,” “bacterial species” and “animal species,” “body site from which bacteria were isolated,” and “AST results from bacterial species that only cause disease in animals” (> 70% extremely comfortable answers, little to no discomfort with sharing). Most participants were also extremely comfortable sharing “AST result data from bacterial species that affect both animals and humans,” “laboratory sample identification number,” and “region of country” or “state where animal was located” (> 50% extremely comfortable answers); however, a level of concern with sharing these items was noted by some participants. From the data items that were not related to geospatial location, the item with the highest level of concern for sharing was “previous antibiotic use” (26.2% somewhat/extremely uncomfortable); however, most participants were still at least somewhat comfortable sharing this information (58.4% extremely/somewhat comfortable). While sharing “the county where animal is located” also showed an overall positive level of comfort (> 50% extremely/somewhat comfortable answers), when geospatial data sharing was more limited and specific, stakeholders demonstrated increasing concern. The item with the most concern noted was “veterinarian’s name and address” (only 20% extremely/somewhat comfortable answers); there was also concern with sharing the “zip code where the animal tested is located,” although not as great (44.6% extremely/somewhat comfortable).

Figure 2
Figure 2

Premeeting survey results on data sharing, listed in order of comfort level (n = 65). AST = Antimicrobial susceptibility testing.

Citation: Journal of the American Veterinary Medical Association 262, 2; 10.2460/javma.23.07.0426

Concerns about sharing AMR and AMU data from animal agriculture—Participants were asked to rank topics in the order of highest concern to least concern regarding sharing of AMR and antimicrobial use (AMU) data from animal agriculture (Table 2). For both questions, the biggest concern was “harm to the public perception of animal agriculture.” Concern regarding FOIA requests was also noted, similar to other answers observed (Table 1). For both questions, the “loss of export markets” was identified as the least concern.

Table 2

Ranking of concerns about sharing AMR (n = 51) and antimicrobial use (AMU; n = 50) data from animal agriculture from premeeting surveys.

Place in the ranking (total points)1 Concerns sharing AMR or AMU data
AMR AMU
1st (257) 1st (272) Harm to the public perception of animal agriculture in general
2nd (255) 3rd (243) FOIA requests
3rd (248) 5th (225) Personal farm liability
4th (243) 4th (238) Risk of increased government restrictions for use of antibiotics in animal agriculture
5th (239) 2nd2 (245) FDA regulatory action
6th (230) 6th (201) Risk of compromise of data that may contain personal identifiers
7th (221) 2nd2 (245) Harm to the public perception of my commodity group
8th (143) 7th (131) Loss of export markets

1Total score and final ranking were calculated by multiplying the number of people who chose that option by weights. For example, if the question had 4 options to be ranked, the total number of people who chose first place was multiplied by 4, the number corresponding to second place was multiplied by 3, third place was multiplied by 2, and the last position by 1. At the end, those numbers were ordered to a final ranking.

2Tie.

FOIA = Freedom of Information Act.

Additional questions—Specific questions to livestock owners and non–livestock owners were also asked (Supplementary Table S2). Although there was a low participation rate from livestock owners, a considerable number of responses were collected from non–livestock owners (n = 59) and revealed that 83% either strongly or somewhat agreed that “[an AMR] database could help producers, animal owners, and veterinarians to detect more rapidly outbreaks of animal pathogens,” indicating support for the use of an AMR database for this purpose. Moreover, all participants were asked whether they would be supportive of efforts to develop statutory data protections for personally identifiable information (PII; somewhat analogous to the Health Insurance Portability and Accountability Act [HIPAA])9 for AMR and AMU data from animals (Supplementary Table S3). A total of 65 responses were obtained, and the percentage of answers that were “maybe” and “yes” combined was 92.3% for AMU and 86.2% for AMR, respectively, denoting that most participants approve the establishment of such efforts.

Virtual and in-person focus groups

Participants’ demographics

From a total preregistration pool of 22, 9 participants attended the in-person focus group, which lasted 3 hours. Participants were asked to select all options that applied, and they self-identified as veterinarians (7), government employees (3), livestock producers (3), others (3), academicians (2), and animal health professionals (1) (Supplementary Table S4). For the virtual focus groups, 49 out of 90 persons who registered participated in the 9 sessions, and based on which focus group they attended, the roles of participants were as follows: researchers (20), government employees (13), VDL representatives (8), and data providers (8) (Supplementary Table S5).

Responses associated with AMR

NVivo analysis of the open-ended responses identified perceived benefits and concerns associated with a centralized AMR database and dashboard. A total of 247 statements were related to benefits and then divided into 14 themes (Table 3; Supplementary Material S5). Regarding perceived concerns, the analysis identified 161 comments that were then divided into 7 themes (Supplementary Material S6).

Table 3

Results of NVivo (QSR International) analysis of perceived benefits and concerns associated with development of a centralized AMR database and dashboard in veterinary medicine.

Themes Description %1
Benefits
   Resource of information and AMU guidance Provide information and contribute with accurate data; support and guide AMU and combine clinical response with AMR data 26.0%
   Surveillance, trends, and emerging AMR Increase ability to monitor trends, understand changes over time, identify emerging issues and genes 24.0%
   Link to One Health components Support animal, human, and environmental health and make connections 9.7%
   Provide centralized data and harmonization Central place to find harmonized and accurate data 8.5%
   Enable comparisons Help identify linkages and compare pathogens, animal species, veterinary services, and industries 6.3%
   Monitor and direct interventions Monitor outcomes of mitigation efforts and legislation and guide policy generation 6.3%
   Improve microbiology references Improve MICs, BPs, AST, ECOFFs, CLSI 5.3%
   Identification of geographic distribution Better understand geographic distribution of AMR, compare regions, and identify target areas 3.6%
   Safeguard future antimicrobial availability Protect access to AM and guarantee its future efficacy 2.9%
   Beneficial for industry and government Pharmaceutical and animal health industry can use data to develop products and identify other needs; government could improve communication and alerts 2.4%
   Link between AMU and AMR Ability to establish connection between AMU and resistance and link with treatment outcomes 1.9%
   Development of antibiograms Creation of antibiograms to be region specific 1.7%
   Parity with other countries Comparison and correspondence between US and other countries 0.7%
   Encourage collaboration Help establish collaboration among stakeholders 0.5%
Concerns
   Unintentional misuse and misinterpretation Incorrect interpretation of data, skewed data 22.6%
   Malicious use of data and fear of government’s actions Retaliation, lawsuit, or that one could use data against operations/individuals; government setting rules about AM use 21.5%
   Data privacy and security To whom data will be available and how it will be protected 17.7%
   Accuracy and standardization Data standardization, validation, and curation to guarantee data accuracy; worries about biased data from VDLs 15.5%
   Lack of data Not enough data due to providers’ hesitancy in participation and other difficulties; skewed data due to lack of representativeness; no breakpoints to classify isolates as resistant 14.3%
   Cost, time, and resources Cost of maintaining the data, time VDL will need to invest, other resources needed 4.9%
   Need of training and education Lack of training and education impacting use of the AMR dashboard 3.4%

1Calculated based on the number of comments and supporters (ie, participants’ use of the Annotate function demonstrating agreement with statements) of each theme and compared to total number of statements categorized as benefits or concerns.

AST = Antimicrobial susceptibility testing. BP = Breakpoint. CLSI = Clinical and Laboratory Standards Institute. ECOFFs = Epidemiologic cutoff values. VDL = Veterinary diagnostic laboratory.

Benefits—The top 2 themes (Table 3) corresponded to 50% of all opinions related to benefits. The first theme was that an AMR dashboard could serve as a valuable “resource of information [on AMR] and AMU guidance” and included comments on helping establish dosing regimens, filling research gaps, correcting assumptions, identifying risk factors and drivers of resistance, and enabling better treatment options. Also, participants commented that such a tool would contribute to obtaining more accurate AMR data to support and guide AMU and help gather funding to improve animal health. The second most cited them was “surveillance, trends, and emerging AMR,” where participants noted that such a tool would increase the ability to monitor AMR trends, recognize patterns, understand changes over time, and identify emerging issues. The possibility of tracking emerging resistance genes was also mentioned. A predetermined list of potential benefits that a centralized AMR repository could provide to veterinarians, producers, researchers, and government was also presented to participants to assess their level of agreement with each topic. The results indicate that most participants agreed there are numerous clear benefits to a centralized AMR repository (Supplementary Table S6).

Concerns—The top 3 themes (Table 3) corresponded to 61.8% of all concern-related opinions and included “unintentional misuse and misinterpretation,” “malicious use of data and fear of government actions,” and “data privacy and security”; some examples of statements include fear of data being inappropriately interpreted, fear of having biased data and skewed results depending on where the data are coming from or due to the lack of representativeness, and concerns about how use of breakpoints in veterinary medicine could lead to a misinterpretation of resistance due to differences between human and animal breakpoints. Also, within those main themes, concerns about data privacy and security, suffering retaliation, lawsuits, and having data used against them, in addition to government setting inconvenient rules about AMU based on data collected, were mentioned.

Participants were then asked to provide comments regarding alleviation of their concerns, and 3 main themes emerged, corresponding to 52.2% of all concern-alleviation statements (Supplementary Table S7). The most mentioned theme was “education and training” (21.0%), which included the need of tutorials, guidelines, and training of providers and users in addition to the need of increasing knowledge of the public, data users, and providers. “Privacy and security” (19.3%) was the second most mentioned theme, indicating that participants would like to have guaranteed confidentiality to avoid repercussions. The third was “standardized, curated, and accurate data” (11.9%), which reflected the need of data curation and harmonization to provide accurate information (Supplementary Material 7).

Geospatial data—A question related to the level of geospatial data participants were willing to share versus what is needed to reap the desired benefits was asked in the virtual focus groups only. While most participants of virtual focus groups agreed that the level of geospatial data needed to provide benefits would be county- or zip code–level data, most participants were not comfortable sharing data below the state level (Supplementary Table S8). Also, participants in virtual focus groups pointed out that the level of geospatial data needed and available depends on the animal species and type of industry. It was frequently mentioned that in some regions of the country, some counties are bigger than some states, so regional differences might affect an individual’s preference for county level if the county is over a certain size. Some participants also commented that “VDLs do not often have this kind of information” and that therefore gathering geospatial data may represent additional steps that will have to be undertaken by the VDLs.

A complementary question about level of comfort on sharing specific geospatial data was asked in virtual and in-person focus groups (Supplementary Table S9). The proposed use of 3-digit zip codes (ie, aggregated at the level of the first 3 zip code digits) for data analysis was discussed, as it is currently considered an acceptable practice for the Safe Harbor method of de-identification of human health-care statistical data under HIPAA.9 Given this information, most participants were quite comfortable sharing a 3-digit zip code (62.6% extremely/somewhat comfortable answers). However, some stakeholders stated a few concerns (eg, “It might not give valuable information,” “There still may be a potential risk of site reidentification depending on size of the operation and kind of commodity,” and “There could be differences between environments within a 3-digit zip code area”). Concerning 5-digit zip codes, participants were shown a deidentified heat map (Supplementary Figure S1) to demonstrate that data could be collected and used without direct association to the actual data. Given this information, attendees agreed that this would be much preferred compared to generation of actual maps (70.3% extremely/somewhat comfortable answers); however, some remained concerned with data security and reidentification risks, and others questioned whether the information would be useful if there is no map or scale associated.

Additional questions—Finally, participants were also asked to select the top 3 data types that they were least comfortable sharing (during virtual focus groups) or any item that they were either somewhat or extremely uncomfortable sharing (during in-person focus group; Supplementary Table S10). Results were similar to pre-meeting surveys and identified that “laboratory sample number” and “previous AMU” were the primary items of concern. In addition, this set of questions specifically asked about sharing “genetic test results” and revealed that WGS was the third least preferred data to be shared because its clinical and public health meaning could be misinterpreted, leading to erroneous linkage to outbreaks.

Postmeeting surveys

Participant demographics

From the 58 stakeholders that participated in the in-person and virtual focus groups, 35 completed at least a portion of the postmeeting surveys (60.3% completion rate). Overall, participant demographics were similar to those reported in the premeeting surveys and represented a wide range of stakeholders (Supplementary Table S11). Participants were asked to select all that applied, and they self-identified as veterinarians (19), academicians (19), government employees (11), others (4), animal health professionals (3), human health sector (3), and livestock producer (1) and represented all sectors of animal agriculture listed. Only 1 participant self-identified as a livestock producer; however, 4 affirmed that they own livestock.

Responses associated with AMR

The postmeeting survey was designed to focus on assessing the changes in the perception related to the topics discussed during the focus group. Therefore, attendees were specifically asked if their perception of benefits and concerns associated with a centralized repository of AMR data from animals had changed. While more than 63.6% responded that their “opinions about concerns had not changed” and 76.5% kept the “same opinion on benefits,” 23.5% stated that they are “more likely to believe that there are significant benefits to this project,” and 36.4% mentioned that they are “less concerned overall about the risks to themselves or industry.” No participants stated that they were “less likely to believe in the benefits” nor that they were “more concerned about the risks.” Some participants added written comments that their opinion has not changed because they already saw benefits and minimal concerns or that, after the focus groups, they had fewer concerns because discussing the positive aspects made them realize the importance of such an initiative. Additional comments were centered around concerns associated with data confidentiality/privacy and security, data misuse, and the need of harmonization to provide an accurate and reliable database.

Attendees were then asked to report their level of comfort on sharing specific data related to AMR, and answers obtained in premeeting surveys were compared with answers from postmeeting surveys (Table 4). Comfort perception improved in all areas except for 3 items: “county where animal was tested” maintained almost the same percentage, while “laboratory sample identification number” and “zip code in which the animal tested is located” had the level of comfort decreased. Sharing zip code information was a concern highly mentioned in all focus groups and surveys due to the perceived risk of tracing AMR source back to farm/individual. Concerns regarding sharing of laboratory sample identification number and risk of reidentification of site of origin were shared in many of the focus groups, which may have influenced postmeeting perceptions. While decreased, the data show that more than 50% of people still felt some level of comfort sharing this data.

Table 4

Comparison of comfort perception in sharing of specific data items between premeeting (n = 65) and postmeeting surveys (n = 32).

Data items Pre- or postmeeting Extremely comfortable Somewhat comfortable Neither comfortable nor uncomfortable Somewhat uncomfortable Extremely uncomfortable
Body site from which the bacteria were isolated Pre- 53 (81.5%) 8 (12.3%) 4 (6.2%) 0 0
Post- 30 (93.8%) 2 (6.3%) 0 0 0
Animal species from which the bacteria were isolated Pre- 50 (76.9%) 11 (16.9%) 4 (6.2%) 0 0
Post- 28 (87.5%) 4 (12.5%) 0 0 0
AST result data if it is from a bacterial species that only causes disease in animals Pre- 46 (70.8%) 13 (20%) 6 (9.2%) 0 0
    Post- 28 (87.5%) 4 (12.5%) 0 0 0
Bacterial species isolated from an animal Pre- 50 (76.9%) 11 (16.9%) 3 (4.6%) 1 (1.5%) 0
Post- 27 (84.4%) 5 (15.6%) 0 0 0
AST result data if from a bacterial species that causes disease in both humans and animals Pre- 43 (66.2%) 11 (16.9%) 4 (6.2%) 6 (9.2%) 1 (1.5%)
    Post- 25 (78.1%) 7 (21.9%) 0 0 0
The region of the country in which the animal that was tested is located Pre-1 42 (65.6%) 12 (18.8%) 4 (6.3%) 4 (6.3%) 2 (3.1%)
    Post- 25 (78.1%) 5 (15.6%) 1 (3.1%) 0 1 (3.3%)
The state in which the animal that was tested is located Pre- 35 (53.8%) 17 (26.2%) 5 (7.7%) 5 (7.7%) 3 (4.6%)
Post- 21 (65.6%) 7 (21.9%) 1 (3.1%) 1 (3.1%) 2 (6.2%)
Any previous use of antimicrobials in this animal that is reported in the case submission Pre- 29 (44.6%) 9 (13.8%) 10 (15.4%) 10 (15.4%) 7 (10.8%)
    Post- 18 (56.3%) 10 (31.3%) 2 (6.3%) 1 (3.1%) 1 (3.1%)
The county in which the animal that was tested is located Pre- 27 (41.5%) 12 (18.5%) 10 (15.4%) 7 (10.8%) 9 (13.8%)
Post- 10 (31.3%) 9 (28.1%) 5 (15.6%) 5 (15.6%) 3 (9.4%)
Laboratory sample identification No. Pre- 33 (50.8%) 11 (16.9%) 8 (12.3%) 9 (13.8%) 4 (6.2%)
Post- 9 (28.1%) 8 (25.0%) 2 (6.3%) 7 (21.9%) 6 (18.8%)
The zip code in which the animal that was tested is located Pre- 19 (29.2%) 10 (15.4%) 14 (21.5%) 7 (10.8%) 15 (23.1%)
    Post- 3 (9.4%) 8 (25.0%) 7 (21.9%) 7 (21.9%) 7 (21.9%)
Any additional testing, including genetic or whole genome sequencing, that was performed on the isolate Pre-2
    Post- 20 (62.5%) 6 (18.8%) 3 (9.4%) 2 (6.3%) 1 (3.1%)

1This question had n = 64.

2Question not asked in premeeting surveys.

AST = Antimicrobial susceptibility testing.

Discussion

Several different approaches were used in this study to collect input from as many stakeholders as possible from a variety of backgrounds. Within the open-ended questions, the idea of using a centralized AMR dashboard as a resource of information and AMU guide was emphasized by participants as well as the ability to monitor trends in AMR, recognize patterns, and identify emerging issues. Within the multiple-choice questions, most participants agreed that all potential benefits proposed to them were clear benefits of this tool, strengthening its importance. Further, pre- and postmeeting surveys revealed that most stakeholders recognize that AMR is a serious concern for human and animal health and that sharing veterinary AMR data from across the country using a centralized database is an important step in combating the development of future AMR. Findings presented by Innes et al10 after interviewing 31 animal agriculture stakeholders within the US were similar to those of our study, reporting that most participants acknowledged AMR as a problem that impacts One Health sectors. A study11 consisting of 39 Tennessee beef cattle producers also utilized focus groups to assess their perception regarding AMR. Similar to our study, those authors reported that participants recognize that AMR emergence is a problem challenging animal and human health. Producers in that study expressed support for more investment in AMR and AMU by federal agencies, reinforcing the need of developing tools such as a central AMR database and dashboard.

While the literature includes studies regarding attitudes associated with AMR in general, there is a lack of reported opinions from stakeholders specifically about sharing AMR data from animals into a centralized database, and this fact was one of the drivers of this work. However, there is a growing body of literature surrounding producer attitudes associated with sharing of agricultural data in general. Producers are concerned about unauthorized access to their farm data, which might lead to reidentification and reputation loss.12 A lack of transparency surrounding issues such as privacy, trust, and liability is also a major factor in reluctance of producers to engage in data sharing.13 These findings align with the concerns noted in our study as well.

In respect for general concerns, incorrect interpretation and inappropriate use of AMR data were mentioned repeatedly by our stakeholders. Other common themes of concern were fear of suffering retaliation, harm to public perception of animal agriculture, and lack of participation and standardization leading to the generation of inaccurate data. Data confidentiality and its protection were also included as significant concerns. When asked about comfort on sharing specific data items related to AMR, items that could potentially lead to identification of data origin (laboratory sample identification number, veterinarian full address/name, and WGS) or have potential risk of misinterpretation (WGS and list of previous use of antimicrobials in the animal) were the ones stakeholders were least comfortable sharing. Participants also voiced hesitancy to support a model (ie, benchmarking) that would allow producers or veterinarians to accuse others, which could also be a reason that discourages them to share a veterinarian’s address. Regarding geospatial data, the more granular the data, the higher the level of concern; this concern was associated with fear of traceback and liability. In addition, participants of the in-person focus group also pointed out that from the production side (ie, swine, dairy, beef), animals move across state and county lines frequently during their production cycle, which makes interpretation of location data very difficult. When stakeholders were encouraged to provide ideas on how concerns could be alleviated, most responses centered on guaranteeing data confidentiality, security, and accuracy. Alleviation of those concerns involves requirements like protection from FOIA, use of aggregate deidentified data only, and avoidance of collecting PII. Premeeting surveys also revealed that, although most participants approve efforts to develop statutory data protections for PII, nonsupportive comments indicated that additional stakeholder feedback related to these issues may be warranted before pursuit of legislation and implementation.

Demands for data curation and standardization, in addition to education, training, and establishment of rules for users and providers, were frequently mentioned to ensure data accuracy. Furthermore, ease of submission of data by VDLs, integration of VDLs in the planning process, transparency, and a plan for frequent communications to increase participation were all recommended. The VDL group also voiced the need of having a better working relationship with human public health and more data integration between sectors, as in the past this working relationship has been strained; while participants believe that there is a need to share the data, these previous interactions have led to concerns over sharing data in the future.

While we attempted to reach as many diverse stakeholders as possible, our results are not without limitations. For instance, although we were able to engage several diverse stakeholders, the participation rate was less than would have been ideal. It is possible that those participants who already believed that AMR was a concern for animal health were the most likely to participate in our study, as participation was entirely optional. Consequently, a selection bias could possibly increase the number of supportive statements and responses to such an initiative. In hindsight, several questions ideally would have been asked in a different format or asked more consistently across all focus groups. For example, the single question on WGS results does not clearly capture the opinions of stakeholders other than indicating significant concerns regarding potential traceback and/or inappropriate use of the data leading to liability or other negative effects. The design of this study also necessitated balancing the amount of participant time required to complete the surveys and/or focus groups with the ability to better inform the participant prior to completion. A lengthy discussion of data confidentiality and security measures that could be implemented was not part of any of the surveys or focus groups, which means that participants were not fully aware of all privacy and security measures that could be implemented to protect the data nor of the entire purpose of collecting such data, prior to providing answers.

In summary, this study found that maintaining open communication on data usage, storage, and security as well as involvement and education of data providers, users, and the public will remain key to engage stakeholders and enable development of an AMR database and dashboard tool for domesticated animals.

Supplementary Materials

Supplementary materials are posted online at the journal website: avmajournals.avma.org

Acknowledgments

Technical assistance during the virtual meetings was provided by Juliana Ruzante, RTI International.

Disclosures

The authors have nothing to disclose. No AI-assisted technologies were used in the generation of this manuscript.

Funding

This work was supported by USDA APHIS.

References

  • 1.

    Antimicrobial resistance fact sheet. WHO. November 17, 2021. Accessed November 30, 2022. www.who.int/mediacentre/factsheets/fs194/en/

  • 2.

    Prestinaci F, Pezzotti P, Pantosti A. Antimicrobial resistance: a global multifaceted phenomenon. Pathog Glob Health. 2015;109(7):309-318. doi:10.1179/2047773215Y.0000000030

    • Search Google Scholar
    • Export Citation
  • 3.

    Ruzante JM, Harris B, Plummer P, et al. Surveillance of antimicrobial resistance in veterinary medicine in the United States: current efforts, challenges, and opportunities. Front Vet Sci. 2022;9:1068406. doi:10.3389/fvets.2022.1068406

    • Search Google Scholar
    • Export Citation
  • 4.

    WHO, Food and Agriculture Organization, World Organisation for Animal Health, UN Environment Programme. Strategic Framework for Collaboration on Antimicrobial Resistance—Together for One Health. WHO, Food and Agriculture Organization of the United Nations, and World Organisation for Animal Health; 2022.

    • Search Google Scholar
    • Export Citation
  • 5.

    Ceric O, Tyson GH, Goodman LB, et al. Enhancing the one health initiative by using whole genome sequencing to monitor antimicrobial resistance of animal pathogens: Vet-LIRN collaborative project with veterinary diagnostic laboratories in United States and Canada. BMC Vet Res. 2019;15(1):130. doi:10.1186/s12917-019-1864-2

    • Search Google Scholar
    • Export Citation
  • 6.

    AMR: introduction to APHIS AMR pilot project. APHIS. Accessed September 15, 2022. https://www.aphis.usda.gov/aphis/dashboards/tableau/amr

    • Search Google Scholar
    • Export Citation
  • 7.

    Stewart DW, Shamdasani P. Online focus groups. J Advert. 2017;46(1):48-60. doi:10.1080/00913367.2016.1252288

  • 8.

    Bingham AJ, Witkowsky P. Deductive and inductive approaches to qualitative data analysis. In: Vanover C, Mihas P, Saldaña J, eds. Analysing and Interpreting Qualitative Data: After the Interview. SAGE Publications; 2022:133-146.

    • Search Google Scholar
    • Export Citation
  • 9.

    US Department of Health and Humans Services. When can ZIP codes be included in de-identified information? Guidance regarding methods for de-identification of protected health information in accordance with the Health Insurance Portability and Accountability Act (HIPAA) Privacy Rule. Accessed December 21, 2022. https://www.hhs.gov/hipaa/for-professionals/privacy/special-topics/de-identification/index.html#zip

    • Search Google Scholar
    • Export Citation
  • 10.

    Innes GK, Markos A, Dalton KR, et al. How animal agriculture stakeholders define, perceive, and are impacted by antimicrobial resistance: challenging the Wellcome Trust’s Reframing Resistance principles. Agric Human Values. 2021;38(4):893-909. doi:10.1007/s10460-021-10197-y

    • Search Google Scholar
    • Export Citation
  • 11.

    Ekakoro JE, Caldwell M, Strand EB, Okafor CC. Drivers, alternatives, knowledge, and perceptions towards antimicrobial use among Tennessee beef cattle producers: a qualitative study. BMC Vet Res. 2019;15(1):16. doi:10.1186/s12917-018-1731-6

    • Search Google Scholar
    • Export Citation
  • 12.

    Kaur J, Hazrati Fard SM, Amiri-Zarandi M, Dara R. Protecting farmers’ data privacy and confidentiality: recommendations and considerations. Front Sustain Food Syst. 2022;6:903230. doi:10.3389/fsufs.2022.903230

    • Search Google Scholar
    • Export Citation
  • 13.

    Wiseman L, Sanderson J, Zhang A, Jakku E. Farmers and their data: an examination of farmers’ reluctance to share their data through the lens of the laws impacting smart farming. NJAS Wagening J Life Sci. 2019;90-91(100301):1-10. doi:10.1016/j.njas.2019.04.007

    • Search Google Scholar
    • Export Citation
All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 19996 19996 618
PDF Downloads 664 664 37
Advertisement