An expert opinion process to prioritize One Health information needs during a zoonotic disease outbreak

The prioritization exercise was conceived in the fall of 2022 at the weekly meetings of the ECCC AIV Oversight and Coordination Committee, a committee consisting of management and working level staff from the Canadian Wildlife Service (CWS), the Science and Technology Branch (STB), Enforcement Branch, and the Public Affairs and Communications Branch that are engaged on the HPAIV file. An ECCC Core Prioritization Project Team was created to refine the specific objectives and ground the prioritization process in One Health concepts. The Core Team members (Giacinti, Wilcox, Sharp, Soos, Ronconi, Leafloor, Duffy, Brown, Avery-Gomm, Provencher) were selected to be representative of perspectives from ECCC personnel spanning geographic regions, jurisdictions, and areas of specialization (e.g., bird health, epidemiology, population monitoring, avian tracking, and migratory bird management), and were integral to collaboratively designing and implementing a process that was informed by multiple branches of ECCC. Starting in November 2022, this team met regularly to discuss the process, design/refine materials and instruments, review documents, and provide feedback. The lead author (Nalepa) was charged with designing, implementing, and analyzing the process with the team's inputs, guided on a day-to-day basis by two members of the Core Team (Provencher and Giacinti).

The ECCC Core Prioritization Project Team sought to engage ECCC professionals representing expertise in a variety of domains, as shown in Fig. 1. Various consensus-based approaches were designed partly around practical considerations to encourage as much expert participation as possible in meeting the objectives. As it was important to include experts representing different regional perspectives from the various management regions of ECCC (Atlantic, Quebec, Ontario, Prairies, Pacific, Northern, and nationwide), online activities (e.g., survey, workshops, small group work) scheduled over the course of several months accommodated experts dispersed across the country, some in remote regions where connectivity can periodically be an issue (e.g., CWS northern offices). Multiple activities allowed for more than one entry point to participation. A flexible timeline also allowed the ECCC Core Team to be as inclusive as possible by allowing partners to recommend other experts from their professional network that might offer valuable insight by participating in one of the surveys (i.e., snowball sampling).

To apply a One Health lens (i.e., public health, domestic animal health, wildlife health) to understanding HPAIV-related knowledge gaps and information needs, the ECCC Core Team aimed to replicate part of the process (i.e., an online survey) with external partners. To this end, the ECCC Core Team engaged with the AIV Interagency Wild Bird Surveillance Committee (AIV IWBSC), which is chaired by CWHC, to create an Interagency advisory group (Dawe, Giacinti, Goldsmith, Harms, Himsworth, Joly, Leonard, Massé, Ronconi, Sharp). Interagency advisors were volunteers from the AIV IWBSC representative of public health, wildlife health, and animal health. Interagency advisors collaborated by helping the ECCC Core Team to adapt the relevant ECCC group process materials to assess expert opinion in their broader network of One Health Partners shown in Fig. 1, including federal partners, provincial/territorial governments and academia. See Supplementary material for a more detailed schematic of the stakeholders involved and the relationships between them (Fig. S1).

To design the surveys for the internal ECCC group of experts (CWS and STB) and the interagency One Health partners, the ECCC Core Team and Interagency advisory group consulted with their respective colleagues to produce a vetted, refined information needs list that reflected their own sector's themes and mandates to be evaluated by their peers in the survey (Obj. 1) and generate a list of suggested criteria that should guide surveillance actions that survey participants could choose from (Obj. 2). The ECCC Core Team blind coded the information needs under five themes that encapsulated the main areas of focus for both cohorts: assessing population impacts in wild birds; disease dynamics; pathogen transmission; surveillance methods and protocols; and data management, reporting, and communication. These themes were coded for analysis purposes, and needs in the survey were not organized by theme.

Survey items in their entirety are shown in the Supplementary material (Tables S1 and S2). Both the internal ECCC and interagency survey began with a consent page that described the survey objectives, survey content, and how the data would be used (Q1, Supplemental material Tables S1 and S2). The knowledge that was acquired during this process was specific to professional opinion on a matter that the experts were predetermined to have (e.g., through their formal professional roles). The first section of the surveys collected demographic information and included questions to capture respondent diversity, experience, and self-reported expertise in six categories of birds (i.e., waterfowl, seabirds, raptors, upland birds, waterbirds, and landbirds), among others. The demographic information collected varied slightly between the two surveys (Tables S1 and S2).

In the survey sections pertaining to Objective 1, the aim was to assess the perceived priority level of the ECCC or Interagency advisors’ information needs (31 and 27 needs, respectively). We asked the participants to rate each information need on a five-point Likert scale according to perceived priority level (i.e., Essential, High, Medium, Low, Not at all a priority), with priority defined as the intersection between urgency and importance. Table 1 illustrates how we then determined the priority level (point-of-consensus) for each of the information needs as well as the degree of consensus among the participating group by using an approach applied by Robert de Loë to explore complex policy questions via a policy delphi (de Loë 1995).

If considering responses from two adjacent categories on the Likert scale shifted the consensus up a level/levels (e.g., from medium to high), the consensus was recorded as spanning two categories on the scale (Table 1). For example, if an information need was found to be “High” priority at a low, single category consensus level (e.g., 51%) and considering the contiguous “Essential” priority responses (e.g., 39%) moved the consensus level to high consensus (at least 80% of votes spanned these categories), the need was recorded as high consensus and the point-of-consensus recorded as “Essential–High” (E-H). Since the ECCC group completed their survey first, the results (i.e., top priority information needs and most popular criteria) were included as items in the interagency survey alongside the additional information needs identified by the Interagency advisory group. There were a total of 12 information needs common to both surveys.

The participants were also asked to justify some of their ratings in an open-ended comment following each rating to provide context for results and content for subsequent workshop discussions. Respondents were instructed to skip any information need that they felt they could not adequately rate based on their expertise. In the survey section pertaining to Objective 2, the aim was to establish the most popular criteria that the respective cohort thought should be used to guide actions when multiple species are affected by HPAIV. Survey participants could choose five from the predetermined list and were prompted for additional suggestions.

We distributed the online survey, which consisted of 44 questions to 65 participants in ECCC's CWS and STB groups and included both scientists and managers. The survey was open to participants from 16–26 January 2023. The interagency survey consisted of 39 questions and was distributed on 30 March 2023 to 62 potential participants. Participants of the interagency survey were composed of AIV IWBSC members and respondent referrals to ensure that a range of diverse perspectives under the One Health umbrella were represented. To accommodate some members of the AIV IWBSC that were not able to complete the survey during the original distribution period, we re-opened the survey for 3 days, which allowed for these additional entries.

Workshop 1 was held virtually and open to all participants of the internal ECCC survey. The overall aim was to reach consensus on the five information needs that did not achieve consensus on a priority level during the course of the ECCC internal survey. For each of those needs, we presented workshop participants with anonymous rating justifications that had been entered into the survey by their colleagues. We presented rationales that offered unique insight, or served as a counter-perspective to rationales of those participants that rated the same need at a different priority level. Participants were allowed approximately 5 min to consider the evidence and bring any additional points to the group either verbally or via the online chat (both methods attributable) for consideration before the participants anonymously re-rated the information need.

The aim of Workshops 2 and 3 were to allow experts to evaluate to what degree the top priority information needs could be met with current resources. The participants were presented with the information needs that reached any consensus level (i.e., low, medium, high) to be Essential–High priority (Workshop 2) or High–Medium priority (Workshop 3) and were prompted to identify perceived challenges or resource gaps associated with getting those information needs met. Using Google Jamboard, a digital interactive whiteboard, we prepared a slide for each information need to be rated for feasibility over the course of the workshop (Fig. S2). Participants were provided prompts to help them identify challenges in resource areas such as human resources, financial needs, technology, logistics as well as an “other” category to include prerequisite knowledge or other issues that required resolution. Participants anonymously added digital sticky notes as they thought of challenges or wanted to support a note that another colleague had already contributed to the Jamboard by adding one that stated “Agree” (Fig. S2). After 5–10 min, participants were asked to anonymously rate each need according to perceived feasibility using the online voting platform in Microsoft Teams using the following Likert scale:

The results from this online feasibility voting were analyzed using the same consensus and point-of-consensus method as described in Table 1. We also used the Google Jamboard slides to identify the main resource area/s that require attention for meeting each need as indicated by areas with a disproportionate number of challenges listed under them or whether many participants upvoted a particular challenge. For example, the main resource area hindering progress in meeting the need “Evaluating the effectiveness of removing carcasses to minimize HPAIV spread among migratory birds” was determined to be primarily logistical in nature (Fig. S2).

Workshop 4 consisted of an exercise to help guide the prioritization of Essential–High information needs that were perceived to be currently under-resourced based on results from Workshop 2 (i.e., needs that were ranked by consensus to be only Fairly feasible or span the Fairly feasible to Less feasible categories on the rating scale). Participants were presented with a summary of the challenges that the group had already identified in Workshop 2 as affecting the feasibility of meeting each need. Participants reviewed these summaries, and after 20 min of open discussion (verbally and in the chat), they anonymously ranked the needs by reflexively answering the question: “If you have limited resources to put toward these Essential–High information needs, what order of priority would you support them in”? Overall rankings were calculated by averaging the individual rankings for each need.

In the last workshop (Workshop 5), ECCC participants were asked to identify program requirements to address the Essential–High information needs. These requirements were then inductively coded into program areas or “leverage points” crucial for determining progress toward fulfilling an information need. This built on results from previous workshops; participants referenced result summaries from Workshop 2 (i.e., the feasibility ratings and main challenges), and broke into small groups based on interest and (or) expertise to identify activities not included in current work plans that would support the fulfillment of the information needs and entered them into a spreadsheet. The participants worked together for approximately 30 min, although some individuals dedicated additional time post-workshop. Coded program areas included: logistics and coordination; field needs; field needs and modelling capacity; project scoping and collaborative planning; and analytical and lab capacity. If the majority of activities listed under an information need fell into the area of “field needs”, for example, this area was considered the current “leverage point”.

The ECCC survey response rate was 74%. Sixty-nine percent of respondents were from CWS, 23% from STB, and 8% non-reporting. Respondents’ average number of years in their current position was 8.5 years (median = 5.5 years), and the average time spent at ECCC was 10 years (median = 9 years). Respondents' work was focused mainly in the Atlantic region followed by Ontario and Quebec (Fig. S3). Self-reported expertise for six bird groups varied across taxa (Fig. S4). Expertise at any level was highest for waterfowl followed closely by waterbirds and seabirds. There was a higher proportion of respondents reporting a high level of expertise in waterfowl and seabirds relative to the other bird categories (Fig. S4).

The interagency survey response rate was 48%, with the survey distributed to a total of 62 experts from the Interagency Wild Bird Surveillance Working Group. Twenty respondents applied wild bird surveillance in the context of wildlife health, eight in animal health, and two in public health (Fig. S5). Overall, the respondents were almost equally split between federal and provincial organizations (43% and 39%, respectively), with the balance representing academia via the CWHC (18%) (Fig. S5). Similar to the ECCC group, the interagency group represented people from across several regions, including those focused nationwide (40%), Atlantic (26.7%), Pacific (10%), Prairies (10%), Ontario (6.7%), and Quebec (6.7%) (Fig. S6).

Interagency expert participants represented 13 agencies and organizations, including four federal agencies with interests in HPAIV (Canadian Food Inspection Agency (CFIA), Public Health Agency of Canada, Indigenous Services Canada, and Parks Canada). Eight provincial governments were represented, including British Columbia, Alberta, Saskatchewan, Manitoba, Prince Edward Island, Newfoundland and Labrador, New Brunswick, and Nova Scotia, as well as academic experts via the CWHC. Notably, there were no respondents in the interagency group that identified as northern, although the nationwide selection may include people that represent northern interests and programs in addition to other areas (Fig. S6).

Information needs reached the highest level of consensus when two adjacent categories on the Likert scale were considered together, thus no information need remained at a single category consensus. From this, we can infer that there was not much polarization in expert opinion regarding item priority levels. ECCC participants considered an equal number of information needs to be Essential–High (E-H) priority and High–Medium (H-M) priority, although overall, the consensus level was higher for the E-H priorities (Table 2). We also found that no information needs identified were L-NP priority at any consensus level, which is not surprising given that this was a curated list of information needs developed by the ECCC Core team. Initially in the ECCC survey, a total of five information needs did not meet consensus on any level. After Workshop 1, in which participants considered their colleagues’ written rating justifications from the survey for “no consensus” items and re-rated them, three (9.7%) still lacked consensus (Table 2).

Overall, the interagency respondents rated 59% of the information needs to span E-H priority (at any consensus level), and most of the remaining needs (33%) to be of H-M priority (Table 3). The only item to reach a high level of consensus spanning the E-H categories was a need exclusive to the interagency survey, a “Harmonized regional and stakeholder reporting of wild bird AIV surveillance data”. The two needs, “Development of strategic recommendations on how to prioritize sample collections (e.g., when are full necropsies recommended vs. swabs vs. serology?)” and “A better understanding of migratory bird species/populations in Canada susceptible to high mortality and morbidity rates” did not reach a clear consensus (Table 3). As with the ECCC process, no information needs reached a single category consensus; all information needs that reached consensus spanned two categories adjacent to each other on the Likert scale and none were considered to be N-P priority at any consensus level. However, unlike the ECCC cohort, interagency participants did not consider any need to be of M-L priority, and thus comparatively rated their needs to be of higher priority overall (i.e., all at least H-M priority). Group conviction in the priority level ratings was comparable between cohorts; although 13% of the ECCC group's needs reached a high consensus level irrespective of priority level as opposed to about 11% of the needs in the interagency group, the interagency participants had more medium consensus needs when compared the ECCC group, which had an equal number of needs reach low or medium consensus (Tables 2 and 3).

The ECCC and interagency groups’ information needs were organized by theme (Fig. 2). Since the surveys were not identical and had a different amount of information needs under each theme, comparisons between groups at the theme level was not straightforward. However, there were some notable observations.

All five of the information needs under the theme of “Assessing population level impacts in wild birds” were considered of E-H priority for the ECCC group. As established E-H priorities, all the needs under this theme were carried over from the ECCC to the interagency survey where the interagency participants considered them to be less of a priority, collectively (Fig. 2).

The interagency group rated their needs under several themes to be of higher priority when compared to the ECCC group including the themes of “Surveillance methods and protocols” and “Data management, reporting, and communication” (Fig. 2). The interagency group also considered their needs under the “Pathogen transmission” theme to be of higher priority, generally; the interagency group had fewer needs (6) to rate than the ECCC group (13) but considered all of theirs to be of E-H or H-M priority, whereas the ECCC group were more divided. The ECCC group rated 38% of their “Pathogen transmission” information needs to be of M-L priority, and all were related to the evaluation of the effectiveness of several approaches in reducing the spread of HPAIV (i.e., pausing research, closing tourism to or scaring birds off colonies, pausing rehabilitation, and bird banding programs).

In terms of individual needs, of the 12 information needs that were common to both surveys, the ECCC group and the interagency cohort agreed on seven in terms of priority level (all E-H), and of those, three reached the same level of consensus between the groups (Table 4).

The top five criteria that ECCC experts thought should guide disease surveillance and research when multiple species or populations are impacted by HPAIV included Species at Risk, species commonly harvested in the US and Canada, species that may pose risks to the commercial poultry industry, species for which Canada hosts at least 10% of the global population, and species likely to have suffered population impacts due to HPAIV in 2021–2022 (Table 5). Given that some regions had very few respondents, regional differences among experts were difficult to establish, but there were some items of note. For example, the only unanimous criterion in the top five across regions was “Species likely to have suffered population impacts due to HPAIV in 2021–2022” with 79% of all ECCC respondents selecting this criterion. Notably, for northern region respondents, a top criterion included prioritizing species that pose human health risks to those with recognized and affirmed treaty rights, which did not factor in the top five overall or with any other regional group.

ECCC respondents had the opportunity to suggest criteria beyond the predetermined criteria choices in the survey via an open-ended prompt (Table S1, Q13). Suggestions included non-listed species undergoing or recently experiencing population declines that threaten conservation or viability, Indigenous Government-identified priority species, species for which actions to mitigate spread or impacts of HPAIV would be easily or moderately easily implemented (e.g., species where recreational or scientific activities are taking place in or around the colony) and species that demonstrate behaviour that increases transmission risk.

The interagency survey showed that the most popular criteria to guide surveillance beyond 2022–2023 was to focus on species known to be asymptomatic carriers that drive large-scale transmission and spread. In total, 77% of the interagency respondents selected this criterion in their top five, which was not reflected in ECCC's top five criteria (Table 5). By comparison, this criterion ranked 8th in the ECCC online survey. Overall, three of the top five criteria from the ECCC survey were also in the top five criteria for the interagency survey. This included species where population level impacts were likely, species that post a risk to commercial species, and those commonly harvested. Notably, species that pose a risk to human health were also included in the interagency top five, whereas this was 7th on the ECCC ranking. Interagency participants’ additional criteria suggestions included prioritizing surveillance during periods where there are juvenile birds in the population that are immunologically naive and vulnerable to mortality in susceptible species or populations. Also suggested was to prioritize Arctic nesting colonial species or provincial populations of an avian species listed under a provincial Species at Risk Act, among other criteria.

During Workshops 2, 3, 4, and 5, the ECCC group further explored the feasibility of meeting the 22 most highly prioritized information needs to better understand resource needs within existing programs. The aim was to determine not only the main challenges in meeting needs but helping the group triage top needs based on the amount of support they required and establish the importance of the E-H needs that were considered under-resourced relative to one another according to the group.

Figure 3 and Supplementary material Fig. S7 (E-H and H-M needs, respectively) show a synthesis of the data gathered from the ECCC survey and Workshops 1–4 to meet Objectives 1 and 3. Eight E-H and H-M needs were rated either to be Feasible–Fairly feasible (F-FF), Feasible (F), or Definitely feasible–Feasible (DF-F). Although four of the needs did not reach consensus on perceived feasibility (all at the H-M level), none of the remainder were considered solidly DF or Infeasible (I), meaning the group considered the majority of the needs to be capable of being accomplished within a reasonable period of time pending varying degrees of additional resources/support. According to participants, the main challenges to meeting most of the relatively lower feasibility E-H needs (indicated by arrows in Fig. 3) included logistical challenges often in combination with human resource limitations and financial constraints (Fig. 3). When Workshop 4 participants ranked the six lower feasibility E-H needs based on relative importance in the face of insufficient resources, the top ranked “A better understanding of migratory bird species/populations in Canada susceptible to high mortality and morbidity rates” required an influx of resources on multiple fronts (Fig. 3). Participants stressed the need for increasing capacity for collaboration with agencies outside of ECCC (sticky notes placed under the “Other” category in Google Jamboard) as a forefront challenge in fulfilling two E-H needs, including the second-ranked “Viral genome analysis to better understand transmission between species and within colonies”. Logistical challenges were heavily implicated as challenges to progress for the lower feasibility H-M needs (Fig. S7). We found that the leverage point for the majority of the E-H needs were in the program area of field needs, or field needs in combination with modelling capacity (Table S3).