Technology in Palliative Care (TIP): the identification of digital priorities for palliative care research using a modified Delphi method

Background Developments in digital health (describing technologies which use computing platforms, connectivity, software, and sensors for health care and related purposes) has the potential to transform the delivery of health and social care to help citizens manage their own health. Currently, we lack consensus about digital health research priorities in palliative care and lack theories about how these technologies might improve care outcomes. Global palliative care need is expected to increase due to the consequences of an ageing population; therefore, it is important for healthcare leaders to identify innovations to ensure that an increasingly frail population have appropriate access to palliative care services. Consequently, it is important to articulate research priorities as the first step to determine how we should allocate finite research resources to a field saturated with rapidly developing innovations. Aims To identify research priority areas for digital health in palliative care. Methods We selected the digital health trends, most relevant to palliative care, from a list of emerging trends reported by the Future Today Institute. We conducted a modified Delphi process and consensus meeting with palliative care experts to identify research priorities. We used the views of public representatives to gain their perspectives of the agreed priorities. Results One hundred and three experts (representing 11 countries) participated in the 1st Delphi round. Fifty-five participated in the 2nd round (53% of 1st round). Eleven experts attended the final consensus meeting. We identified 16 priorities areas, which were summarised into eight themes. These themes were: big data, mobile devices, telehealth and telemedicine, virtual reality, artificial intelligence, the smart home, biotechnology and digital legacy. Conclusions The identified priorities in this paper represent a wide range of important emerging areas in field of digital health, personalised medicine, and data science. Human-centred design and robust governance systems should be considered in future research. It is important that the barriers and risks of using these technologies in palliative care are properly addressed to ensure that these tools are used meaningfully, wisely and safely and that do not cause unintentional harm.

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

ABSTRACT Background
Developments in digital health ( describing technologies which use computing platforms, connectivity, software, and sensors for health care and related purposes) has the potential to transform the delivery of health and social care to help citizens manage their own health. Currently, we lack consensus about digital health research priorities in palliative care and lack theories about how these technologies might improve care outcomes. Global palliative care need is expected to increase due to the consequences of an ageing population; therefore, it is important for healthcare leaders to identify innovations to ensure that an increasingly frail population have appropriate access to palliative care services. Consequently, it is important to articulate research priorities as the first step to determine how we should allocate finite resources to a field saturated with rapidly developing innovations.

Aims
To identify research priority areas for digital health in palliative care.

Methods
We selected the digital health trends, most relevant to palliative care, from a list of emerging trends reported by the 'Future Today Institute'. We conducted a modified Delphi process and consensus meeting with palliative care experts to identify research priorities. We used the views of public representatives to gain their perspectives of the agreed priorities.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Results
One hundred and three experts (representing 11 countries) participated in the 1st Delphi round. Fifty-five participated in the 2 nd round (53% of 1 st round). Eleven experts attended the final consensus meeting. We identified 16 priorities areas, which were summarised into eight themes. These themes were: big data, mobile devices, telehealth and telemedicine, virtual reality, artificial intelligence, the smart home, biotechnology and digital legacy.

Conclusions
The identified priorities in this paper represent a wide range of important emerging areas in field of digital health, personalised medicine, and data science. Humancentred design and robust governance systems should be considered in future research. It is important that the barriers and risks of using these technologies in palliative care are properly addressed to ensure that these tools are used meaningfully, wisely and safely and that do not cause unintentional harm.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. has the potential to transform the delivery of health and social care to help citizens manage their own health. 1-3 Currently, we lack consensus about digital health research priorities in palliative care and lack theories about how these technologies might improve care outcomes. Therefore, it is important to articulate research priorities as the first step to determine how we should allocate finite resources to a field saturated with rapidly developing innovations. Global palliative care need is expected to increase due to the consequences of an ageing population; therefore, it is important for healthcare leaders to identify innovations to ensure that an increasingly frail population have appropriate access to palliative care services. 4 Research demonstrates that, when used well, digital health initiatives improve healthcare delivery and access, 5-15 and the World Health Organisation (WHO) promotes that digital health should be an integral part of health priorities as a means to improve health on a global scale. 16 17 To date, many barriers have prevented the meaningful use of digital health in palliative care; 18 these barriers include expense, inter-operability issues, data privacy and security concerns, lack of effectiveness, equity, and the concern that technology will reduce face-to-face consults between patients and clinicians. 19 20 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

Identification of technology trends from the Future Today Institute
We selected technology trends most relevant to palliative care from a list of emerging technology trends reported by the Future Today Institute (FTIhttps://futuretodayinstitute.com). The FTI is a multi-professional organisation that uses data-driven applied research to develop models that forecast risk and opportunity across several disciplines, which are mapped into technology trends.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

Selection of technology trends for palliative care
We developed criteria to select the FTI trends, based on recommendations from a UK-based policy report, which reported public and professional views on new types of healthcare data. 29 We developed the following statement to select FTI trends for inclusion: 'Trends should involve analysis or use data generated by a patient, caregiver or healthcare professional with potential use in palliative care'. Two authors (ACN and TMc) reviewed all 225 FTI trends. We included 95 (42.2.%) of the trends. We then combined and simplified similar trends to reduce the number to 32 ( Figure 1 -Flow diagram to outline study process for identifying research priority areas). To ratify the validity of the trends for palliative care, we conducted a focused literature review to identify examples where these technologies have been used in healthcare. We used an Excel spreadsheet to collate this data for reference. We developed 32 items for inclusion in the Delphi questionnaire, which reflected the 32 trends identified from the Future Today Institute Report (see Figure 1 -Flow diagram to outline study process for identifying research priority areas). We used Google Forms (https://www.google.co.uk/forms/about/) to develop the survey. We designed the questionnaire to collect demographic information (geographic location, . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 27, 2021. ; https://doi.org/10.1101/2021.06.24.21259307 doi: medRxiv preprint 8 age, occupation), and individuals' rating of importance for each item via a 5-point Likert scale (1 = low priority, 5 = high priority). To ensure that the survey questions were appropriate, we conducted a local pre-study pilot of the questionnaire and supporting materials (Appendix -'Delphi Questionnaire' and 'Scoping review').

Participant Recruitment and Consent
We solicited a convenience sample of palliative care professionals with expressed interest in technological innovation; we used professional networks, social media and email to contact individuals (Appendix -Summary of the networks used to invite palliative care professionals to participate). Consenting participants accessed the study material online to complete an electronic consent form and the 1st round Delphi questionnaire. We invited participants who completed the 1 st round questionnaire to participate in the 2nd round.

Ethical approval
This study was approved by the University of Liverpool Ethics Committee (study approval number 3564).

Data collection and analysis
Quantitative statistical analyses of participants ratings were undertaken with the statistical software package SPSS 22.0. We used the Interquartile Range (IQR) to determine the level of agreement on the five-point scales for each 'area' on the questionnaire. The justification for the levels of agreement were based on thresholds previously used in palliative care Delphi studies, which used a 5-point Likert scale to determine agreement (Appendix: Interquartile Range to be used to guide the Level of . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 27, 2021. ; https://doi.org/10.1101/2021.06.24.21259307 doi: medRxiv preprint 9 Agreement for Delphi responses). 22 30 We emailed a summary of the 1 st round Delphi results to each participant. The email included the following information: (i) a summary of how the participant rated each item in the first Delphi round, and (ii) a summary of all participants' responses for each item (pooled level of agreement). We provided this information so participants could consider whether they wished to rank items differently in the 2nd Delphi round, based on the ranking data generated by other participants.

Round 2 Delphi questionnaire
We provided participants with an electronic link to access the 2 nd round Delphi questionnaire. We asked participants to answer the same questions that were included in the first-round questionnaire. Participants were required to complete the questionnaire within 4 weeks. We analysed responses from the 2nd questionnaire by IQR to provide a final list of items according to their level of agreement.

Final consensus meeting and voting
We organised a consensus meeting to agree the trend list as the final stage of the Delphi process. 22 We invited all participants to attend the meeting at the University of Liverpool, UK. We divided participants into two groups. We attempted to ensure the groups were similar by allocating individuals according to their gender, experience and occupation. We provided participants with the Delphi results, via (i) an oral presentation and (ii) a written summary. ACN and TMc acted as group facilitators and ACN chaired the meeting. We facilitated group discussion and voting to determine which items were included or excluded from the final list. Each item was . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 27, 2021. ; https://doi.org/10.1101/2021.06.24.21259307 doi: medRxiv preprint 1 0 discussed and debated, and a 'raised-hand' vote was undertaken within each group to determine if each item was included or excluded from the final list.
After voting, we compared the outcomes for both groups. We included items if both groups voted for their inclusion. Similarly, we excluded items if both groups voted for exclusion. When the groups disagreed (i.e., one group voting for inclusion and the other voting for exclusion), we facilitated debate with both groups together, which was followed by rounds of voting until consensus was achieved.

Public engagement workshop
Following the Consensus meeting, we conducted a public engagement workshop with lay representatives to determine their views on the agreed priorities. We used volunteer coordinators from Marie Curie Hospice Liverpool and Liverpool University Hospitals NHS Foundation Trust, to invite palliative care volunteers (by telephone and email).

Round 1 Delphi Questionnaire
Round 1 included 103 people participants (Table 1  Level of agreement for each 'priority area' following both Delphi rounds), which suggests that participants considered most items were important.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Round 2 Delphi
Fifty-five (53%) of the round 1 participants completed the round 2 questionnaire. The median age was 44 years, which was similar to round 1. More women than men completed the questionnaire (n = 32, 58.2%). The distribution of occupations was similar across both rounds. Fewer countries (n = 8) were represented among the final sample. The final IQR analysis (Appendix: Level of agreement for each 'priority area' following both Delphi rounds) demonstrates that most items (n = 21, 65.6%) had low levels of agreement, with two (6.3%) and nine (28.1%) items achieving moderate and high levels of agreement respectively.

Consensus meeting and final list of priorities
Eleven people participated in the consensus meeting (10.7% of total participants and 20% of second round participants). The median age of participants was 44, and most were female (n = 7, 63.6%). All participants were based in the UK and were mostly from clinical (n = 6, 55%) or academic backgrounds (n = 4, 36%). The debate resulted in agreement, rejection, modification (rewording and combination) of trends, and the addition of a new item, digital legacy (Appendix: Voting outcomes for consensus meeting). We classified the priorities into eight themes which were: big data, mobile devices, telehealth/telemedicine, virtual reality, artificial intelligence, the smart home, biotechnology and digital legacy ( Table 2 -Final list of priorities).

Public Engagement Event
We conducted the public engagement event at Marie Curie Hospice Liverpool, UK, which was attended by six lay representatives, two staff members (nurse and doctor) and a medical student. In this meeting, we presented the Delphi outcomes and we . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 27, 2021. ; https://doi.org/10.1101/2021.06.24.21259307 doi: medRxiv preprint 1 2 facilitated round table discussions to explore attendees' views on the agreed priorities. We asked attendees to identify areas within these themes that they wanted researchers to study further.
Our public representatives recommended that future research should: (1) ensure a human centre co-design approach to ensure that technologies are designed according to the needs of individuals and (2) that appropriate governance processes should be in place to evaluate efficacy, effectiveness and ethical issues of current and future digital health tools and systems.

Contribution and strengths of this paper
The outcomes of our detailed analysis (involving a modified Delphi process and patient engagement workshop) indicates further digital health research is needed to identify how to best use technology to support palliative care. Our paper is the first priority-setting paper for palliative care digital health and provides a foundation for digital health focused palliative care research.

Mobile devices and wearables
Many studies have described how mobile devices and wearables can support palliative care (e.g. remote monitoring of physical activity and symptoms, to deliver . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) Furthermore, studies should provide more information of how mobile devices can help patients to record their care preferences (e.g., advance care planning). 65 66 Virtual reality Virtual reality (VR) is a human-computer interface technology that uses visual graphics, sounds and other sensory input to create an interactive computer world. 67 Previous studies have described the potential to use VR to support psycho-social symptoms and wellbeing; however, most work is unevaluated so further research is needed. [68][69][70][71] We recognise the potential of VR to support palliative care education; 72 73 however, the Consensus group did not identify this as a current priority. Following our study, we recognize that COVID19 pandemic has accelerated the use of virtual learning environments for medical education, 74 particularly with the potential to use VR for communication skills training. 72 Consequently, it is possible that VR for education would rate higher as a palliative care digital health priority if this study were repeated.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

Biotechnology
Biotechnology involves the combination of technology with living things. 78 Palliative care developments in this area include use biomarkers to predict survival. 79 80 Other studies have identified biomarkers to predict constipation 81

Digital legacy
A digital legacy is the digital information available about someone after death, such as social media profiles, photos, videos and gaming profiles. 87 The volume of digital information generated by citizens is increasing, which creates new challenges after . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. death. 88 The increasing use of cloud storage and social media is contributing to uncertainty of data ownership, which creates difficulties for caregivers to manage the digital legacy of the deceased. Studies demonstrate that healthcare professionals can positively support their patients to manage their digital legacy. 87 89 90 However, digital legacy is not routinely discussed in clinical practice, which means that we generally do not know how individuals want their data to be managed after death. 91 Therefore, we believe that researchers should explore how patients and caregivers can be supported to manage their digital legacy after death, with exploration on the different methods and materials that can be used.

Relation to previous work in this area and areas of interest following the novel COVID19 pandemic
Our study is synergistic with previous work, which has been conducted across the theme areas. 19 33 We acknowledge that our study pre-dates the pandemic and it is possible that the priorities we identified may now have shifted. However, we believe our research findings are valid as the digital health innovations adopted during the pandemic were in sync with our priority list. (Appendix: Examples of technologies used in palliative care during the COVID19 pandemic). 32 33 Telehealth was commonly used during the pandemic, with many palliative care services using this to provide remote clinical support, [92][93][94][95][96][97][98][99][100][101][102][103][104] to communicate 105 and for education. 106 Technologies were used to maintain connection, and to develop communities of palliative care practice. 107 108 VR was used to provide psychological care and symptom management. 109 110 In general, the findings these studies are positive about the potential of digital health to improve palliative care; however, its rapid . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 27, 2021.

Limitations
It is possible that recent developments were not reflected in the priority list due to ongoing advancement of healthcare technologies. For example, the FTI trends list is now in its 2021 version and includes new trends such as, home medical laboratory tests and remote metabolic monitoring. Therefore, it is possible that relevant areas are absent from this analysis. Also, a weakness of digital health research is the rapid change associated with technology, which may cause the findings of this study to lose relevancy overtime.
Our decision to reduce the number of trends from 95 to 32 items, has broadened the focus of the list, which means it is possible that more specific and technical areas were not explored in greater depth (e.g., faceprints, voiceprints, chatbots etc etc). It is also possible that our Delphi participants will have different views on priority of some areas post COVID19, due to the resulting increase of clinical applications of digital health. Questionnaires were mostly completed by participants arising from English-speaking countries, meaning that the experience of non-English speaking populations may not be reflected. It is possible, due to the novel nature of some areas, that participants gave more priority to familiar areas and therefore, less priority to unfamiliar areas. The final priority list may not represent non-UK healthcare systems, as the consensus meeting was only attended by UK residents.
Furthermore, people from different professional backgrounds (including cultures and . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 27, 2021. ; https://doi.org/10.1101/2021.06.24.21259307 doi: medRxiv preprint 1 9 settings) may assign different levels of priority to trends, due their experience, workrequirements and personal beliefs. As most participants were clinically-focused, it is possible that the priorities were orientated to clinical-utility, rather than methodology.

Relevance to research, practice and policy
Decision-makers should ensure that technology is relevant to the needs of the palliative care user, as these requirements will influence the design, use and function of systems. For example, healthcare professionals will generally use technology to access patient data and communicate with other professionals, whereas patients may wish to access their own health data and to contact healthcare services. Further research is needed to develop specific use-cases for these scenarios, to ensure that the technology can be used meaningfully to achieve the intended outcomes.
Furthermore, as the user requirements of people with palliative care needs may differ from the general population 111  collaboratively with creative industries (e.g., designers, developers and engineers) to ensure that designed technologies fulfil the user requirements for specific palliative care use-cases.

Conclusion
The identified priorities in this paper represent a wide range of important emerging areas in the field of digital health, personalised medicine, and data science. Humancentred design and robust governance systems should be considered in future research. Transdisciplinary studies using a range of methodologies are required to further study this priority list. It is important that the barriers and risks of using these technologies in palliative care are properly addressed to ensure that these tools are used meaningfully, wisely and safely and that we do not cause unintentional harm.

COMPETING INTEREST STATEMENT
The authors declare no competing interests.

ETHICS
This study was approved by the University of Liverpool Ethics Committee (study approval number 3564).

DATA AVAILABILITY
The authors declare that the data supporting the findings of this study are available within the paper and its supplementary information files. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

LIST OF ABBREVIATIONS USED
The copyright holder for this preprint this version posted June 27, 2021. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 27, 2021. ; https://doi.org/10.1101/2021.06.24.21259307 doi: medRxiv preprint Monitoring of Cancer Patients in Palliative Care Aiming to Predict Health . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 27, 2021. ; https://doi.org/10.1101/2021.06.24.21259307 doi: medRxiv preprint