Establishing consensus on key elements and implementation enablers of community-based pain programs to support primary health network decision making: an eDelphi study

Keywords: chronic pain, preventive health services, primary health care, secondary prevention.

The global burden of chronic pain is significant and growing. Estimates indicate that by 2050, 5.23 million Australians (16.9%) will be living with chronic pain at a cost of A$215.6 billion, which largely includes productivity costs as a result of the negative impact of chronic pain on individual health and well-being (Deloitte Access Economics 2019). Barriers to reducing the burden of chronic pain include limited access to pain services and limited awareness of multidisciplinary treatment options, with an over-reliance on pain medications (Australian Government Department of Health 2019). There is wide acknowledgement that investing in research and health service improvement in the primary care setting is needed to address these issues (Smith and Torrance 2011; Australian Government Department of Health 2019).

Best practice secondary prevention and management of chronic pain involves a multidisciplinary, biopsychosocial approach to support patient self-management (Kamper et al. 2015). Australia’s primary healthcare system supports patient self-management via community-based chronic disease self-management education programs and financial incentives (via Medicare Benefits Schedule (MBS) items) for health professionals to develop care plans and provide multidisciplinary care (Jordan et al. 2008). While evidence continues to emerge for community-based pain programs (Kawi 2014; Turner et al. 2018; Hurstak et al. 2019; Joypaul et al. 2019), the absence of sufficient MBS support to implement this model of care has hindered the provision of these programs (Jordan et al. 2008; Painaustralia 2019).

Primary health networks (PHNs) are well placed to address this gap in service provision as federally funded commissioning bodies tasked with identifying and responding to local health service needs (Australian Government Department of Health 2016). However, despite widespread service access and opioid prescribing issues, currently only approximately one-quarter (8/31) of PHNs are commissioning community-based pain programs (De Morgan et al. 2019). Recent consultation with PHN representatives found that most were unaware of what other PHNs are doing to support better prevention and management of chronic pain, and that most would be interested in this information and the opportunity to discuss resource and capacity requirements for implementation with other PHNs (De Morgan et al. 2019).

Although there is guidance for program facilitators on the clinical aspects of community-based pain programs (Torrance et al. 2011; Sharpe et al. 2020), guidance for commissioning bodies on the design and implementation of these programs is needed. Commissioning has been recognised as a complex process that requires the commissioning body to establish a clear and evidence-based vision for service design (Robinson et al. 2016). Given commissioning primary care services in Australia is a relatively new responsibility of PHNs (Robinson et al. 2016) and chronic pain programs to date have largely been provided in the tertiary setting (Tardif et al. 2021), drawing on local expertise to establish consensus on the key considerations for the delivery of this model of care in the community would support PHNs to address this service gap.

The aim of this study was to establish expert consensus on best practice key elements of community-based pain programs, as well as enablers important for program implementation and sustainability. These findings will serve as a foundation for research and the synthesis of information to support PHN decision making concerning the commissioning of pain programs.

This study used an adapted method of the traditional Delphi approach to derive consensus known as the ‘reactive Delphi’, where a predeveloped list of items was proposed in the first survey (McKenna 1994). This approach has been preferred over a qualitative first round where a list of items is generated by the expert panel for three reasons: (1) the researchers had compiled a list of key community-based pain program features based on a previous review of primary care chronic pain initiatives that identified Australian pain programs (Walker et al. 2020) and extensive consultation with PHN representatives regarding the design and implementation enablers of initiatives currently being commissioned by PHNs (De Morgan et al. 2019, 2020); (2) to reduce the time commitment of the expert panel; and (3) to minimise the amount of data to consolidate for the second survey so that this research could be completed in a timely manner.

This online Delphi process (eDelphi) involved the commonly adopted three survey rounds (Diamond et al. 2014), which were administered via REDCap (Harris et al. 2009) to a panel identified with expertise in pain programs in Australia. The process aimed to reach consensus on what was originally called a list of principles and a list of enablers of community-based pain programs by iteratively collecting and synthesising expert opinion. Based on expert feedback, this changed to a list of key elements and enablers of community-based pain programs, which is explained in more detail in the study results.

This eDelphi study followed the stages outlined by Mead and Moseley (2001), namely expert panel selection, formulating the questions, statement generation, reduction and categorisation, rating and analysis and iteration, and the reporting guidelines and checklist provided by Hasson et al. (2000) for the design and reporting of Delphi surveys.

Ethics approval to conduct this study was obtained from the University of Sydney Human Research Ethics Committee (Project no. 2020/342) and consent was implied if the survey was completed.

The initial list of best practice principles and implementation enablers was developed for the first survey by drawing on author knowledge from experience working in the field of pain and the previously mentioned literature (De Morgan et al. 2019, 2020; Walker et al. 2020). An initial draft was developed by one author (SDM) and discussed and refined over several meetings with another two authors (PW and FB).

The first survey was piloted by three colleagues of the research team before implementation in line with best practice (Hasson et al. 2000; Powell 2003; Gill et al. 2013), focusing on the clarity and language used in the instruction and questions, survey length and platform accessibility. Only small formatting changes resulted from piloting the survey.

In the first survey, experts were asked to indicate their opinion on the relevance of each of the 10 proposed best practice principles and 22 proposed implementation enablers (categorised into five themes). Experts were asked to indicate whether items should: remain in the list (unchanged), be removed from the list (not relevant to pain programs), be rephrased, be merged with another principle (because they refer to the same thing) or be included in the other list because they were an enabler rather than a principle or vice versa.

Experts could provide a rationale regarding any responses and add any additional (up to five) principles or enablers they felt should be included in each list. Demographic information was also collected in the first-round survey to describe the expert panel, including level of education, their pain-related roles, years of experience, Australian jurisdiction and area (metropolitan/regional).

The second and third surveys used a five-point Likert scale to establish the importance of each item in the list of key elements and implementation enablers. Experts were asked to indicate whether items were considered essential (must be part of community pain programs), very important (should be there but not essential), of average importance (an important part of community pain programs, but its absence will not change the program dramatically), of little importance (nice to have it, but it will not be missed if it is not included) or not important at all (remove from the list). Experts were again offered the opportunity to provide a rationale for their responses or suggest any changes at the end of each list, and any other comments at the end of the survey.

Purposive sampling was used to identify and invite experts to participate in this process. Experts of interest included academics with expertise in pain research, clinicians specialising in pain management, developers and/or commissioned providers of pain programs, executive-level staff members of a peak pain agency and consumer advocates.

All invited participants were based in Australia, were already known to the researchers and represented a range of relevant disciplines (e.g. pain specialists, GPs, psychologists, physiotherapists, pharmacists and nurses). Sixteen experts were invited to participate in line with the literature, which recommends 10–18 experts on a Delphi panel (Okoli and Pawlowski 2004).

This eDelphi process was conducted between June and September 2020. Individual survey links were sent to participants via email through REDCap, except for the third survey, which was sent from one of the researchers with the individual survey link copied from REDCap so that an individualised report could be attached.

Participants received brief feedback regarding changes that had been made to the list of key elements and enablers based on survey results on the first page of Surveys 2 and 3, with the addition of an individualised report to allow participants to review their responses from Round 2 against the average response of the group before completing the final survey (Hasson et al. 2000).

In line with the Delphi approach, participant identifiers were used so that researchers could remind non-responders and seek any necessary clarification on individual responses (Okoli and Pawlowski 2004). Each survey was live for 2 weeks, with a reminder sent out 1 week before the closure date to prompt participants to complete the survey. Subsequent surveys were only sent to those who participated in the previous survey round.

Consensus of 70% agreement was used, as recommended in the literature (Sumsion 1998) and in line with other studies (Suris and Akre 2015; Arblaster et al. 2018; Williams et al. 2018), to retain a best practice key element or implementation enabler in the list (for Round 1, 70% responding ‘it should remain in the list’ or ‘it should be rephrased’; for Rounds 2 and 3, 3.5/5 on the importance rating scale).

Results were analysed using SAS version 9.4 to report on the number of respondents, the mean ± s.d. for responses and the percentage of responders indicating that each principle and enabler should remain unchanged or be rephrased (Round 1) and is essential (5/5 rating) or very important (4/5 rating; Rounds 2 and 3).

The list of best practice key elements and implementation enablers were refined for Surveys 2 and 3 based on the results from the previous survey. This involved one author (PW) synthesising expert feedback and proposing areas for improvement for review and discussion with another two authors (SDM and FB) to come to an agreement on modifications to each list.

The first survey achieved a 69% (11/16) response rate. One of these responses was a nominated proxy in place of an invited expert who also met the inclusion criteria to participate (involved in at least one pain-related role of interest). Only one expert did not go on to complete the second and third surveys due to time commitments. The proportion of experts that felt the survey was very easy or easy to complete increased with each survey (45% for Survey 1, 80% for Survey 2), with all participants finding the survey easy to complete by Round 3. Table 1 presents the demographics of the expert panel that completed all three survey rounds.

Table 1.  Demographics of the expert panel that participated in all three survey rounds (n = 10)

All 10 proposed best practice principles of community-based pain programs achieved at least 70% agreement in Round 1 and remained in the list. Three of the 22 proposed implementation enablers fell below the 70% agreement, but the four (100%), four (80%) and two (50%) experts that did not agree indicated that these three enablers should be moved to the list of principles.

The research team reflected on expert feedback that some enablers were seen to be better suited to the list of principles. Over the three survey rounds it became clear that what were initially thought of as principles were better described as elements of community-based pain programs that focused on what PHNs could practically apply when selecting or designing their own programs. These elements included ensuring program facilitators were trained in pain management, including a pre-program session, incorporating activities in addition to education in group sessions, tailoring programs for different population groups (e.g. Aboriginal and Torres Strait Islander peoples) and items concerning program monitoring, evaluation and continuous improvement.

In addition to moving items between lists, some rephrasing and the addition of new items to the list were necessary to incorporate feedback from the expert group relating to the importance of incorporating behaviour change principles, including family members or carers to support consumers, engaging experienced consumers to help validate participating consumers’ lived experience with pain and to consider using multiple funding streams to support program implementation and sustainability.

The number of changes made to the two lists and the resulting number of items for Survey 2 are detailed in Table 2. This resulted in a list of 18 key elements and 14 implementation enablers, which were grouped under subheadings to break these longer lists into smaller coherent sections and to give a sense of the overall themes of best practice program design and implementation (Tables 3, 4).

Table 2.  Changes made to items proposed in Survey 1

Table 3.  Final list and ratings of importance for each key element from survey rounds 2 and 3 (with 1 being not important at all and 5 being essential) (n = 10)
Unless indicated otherwise, data are presented as the mean ± s.d. CALD, culturally and linguistically diverse

Table 4.  Final list and ratings of importance for each enabler from Survey Rounds 2 and 3 (with 1 being not important at all and 5 being essential; n = 10)
Unless indicated otherwise, data are presented as the mean ± s.d.

All 18 key elements and 14 implementation enablers reached the minimum 70% agreement to remain in the list in both Surveys 2 and 3. The only change made to items included in Survey 2 was slight wording changes to two elements (see Elements 7 and 13 in Table 3) after Round 2 and rewording of two elements (see Elements 3 and 10 in Table 3) after Round 3. These changes were based on feedback from three experts in Round 2 and two experts in Round 3, who were consulted between survey rounds via email and contributed to the rewording of these elements. The three surveys with the three sets of lists that were provided to participating experts are available as Supplementary Material. The final list and mean scores with the percentage of the group that indicated items were either essential or very important from Surveys 2 and 3 for key elements and implementation enablers are presented in Tables 3 and 4 respectively.

This research has demonstrated a high level of consensus among Australian-based experts in the field of pain for the key elements and enablers for implementation of community-based pain programs. The final response rate of 62.5% falls short of the recommended 70% (Sumsion 1998), but this aligns with the recommendation of having at least 10 experts in a panel (Okoli and Pawlowski 2004) and we found no notable difference in the type of pain-related roles of the invited experts that did and did not participate in the study.

The value of this approach, despite opting for the time-saving method of providing predetermined items in the first survey, was realised in the contribution of experts in identifying important additional items. This included consumer-focused items, such as the engagement of consumers in the delivery of programs and the inclusion of family members and carers for consumer support, which emulate priorities reported by consumers in research on international pain programs (Wainwright et al. 2014) and other health promotion programs (Middleton et al. 2016).

Another important enabler identified by experts was the consideration of multiple funding sources to support program implementation and sustainability. Although the benefits of co-design and co-commissioning in the context of service design by PHNs have been identified (Australian Government Department of Health 2018), a recent review of the PHN program has highlighted the need for more government support to build capacity in coordinated commissioning (The Evaluation Team 2018).

The overarching goal of conducting this research is to promote scale-up of community-based pain programs by compiling evidence to support PHN decision making as part of their population health planning and commissioning cycles. However, it must be emphasised that this evidence is one piece of what is required to achieve this goal. As highlighted by Robinson et al. (2016), commissioning is a complex process and requires technical information in addition to critical relationships to work with local providers to tailor programs to the local context. Achieving this goal also relies on PHNs identifying a local need to address the secondary prevention and management of chronic pain, which currently is not the case for all PHNs (Walker et al. 2021).

We also recognise that although the focus of this research is to support PHNs, our findings largely reflect best practice primary health care and could be applied to pain management service delivery more broadly, regardless of funding and delivery arrangements.

Although done purposively to address our aim of supporting PHNs, this research included Australian-based field experts and was specific to the Australian context, and therefore may not be applicable to the design and implementation of pain programs in other primary care settings and in other population settings. We also acknowledge the potential for biased responses or limited resulting items by providing a predetermined list in the Round 1 survey (Hasson et al. 2000).

This eDelphi process has successfully derived consensus among field experts on the key elements of community-based pain program design and enablers important for program implementation and sustainability. The resulting lists will serve as a resource for PHNs considering the commissioning of community-based pain programs and be used in future research working to synthesise information about currently available programs to further support national scale-up and address the increasing burden of chronic pain.

The datasets generated and analysed during the present study are available from the corresponding author on reasonable request.

The authors declare that they have no competing interests.

Funding for this research has been provided from the Australian Government’s Medical Research Future Fund (MRFF). The MRFF provides funding to support health and medical research and innovation, with the objective of improving the health and well-being of Australians. MRFF funding has been provided to The Australian Prevention Partnership Centre under the MRFF Boosting Preventive Health Research Program. Further information on the MRFF is available at www.health.gov.au/mrff. This research was also supported by the Sydney Medical School Foundation, University of Sydney.