Partnering with frail or seriously ill patients in research: a systematic review

Background The expectation to include patients as partners in research has steadily gained momentum. The vulnerability of frail and/or seriously ill patients provides additional complexity and may deter researchers from welcoming individuals from this patient population onto their teams. The aim was to synthesize the evidence on the engagement of frail and/or seriously ill patients as research partners across the research cycle. Methods A systematic review was conducted using PRISMA guidelines. A search strategy included MEDLINE®, EMBASE®, Cumulative Index to Nursing and Allied Health Literature (CINAHL), and PsycINFO from database inception to April, 2019. Eligible studies were peer-reviewed qualitative, quantitative, and mixed methods research reporting on the engagement of frail and/or seriously ill patients as partners on research teams. The Mixed Methods Appraisal Tool was used to appraise study quality. Narrative analysis was conducted. Results Of 8763 citations, 30 were included. Most studies included individuals with cancer on the research team (60%). Barriers included: lack of time and resources (50%), discontinuity in contribution (37%), and concerns for well-being (33%). Facilitators included: trust and mutual respect (60%), structural accessibility (57%), flexibility in timing and methods of engagement (43%), and attention to care and comfort, (33%). Perceived impacts for patients included: renewed personal sense of agency (37%) and emotional/peer support (37%). Impacts for researchers included sensitization to the lived experience of disease (57%) and an increased appreciation of the benefits of patient engagement (23%). Research design, execution, and outcomes, developed with patients, were deemed more suitable, relevant and reflective of patients’ priorities. Conclusions There is emerging evidence to suggest that research partnerships with frail and/or seriously ill patients can be achieved successfully. Patients mostly report benefit from partnering with research teams. Frailty and/or serious illness do present legitimate concerns for their well-being but appear to be successfully mitigated when researchers ensure that the purpose of engagement is well-defined, the timing and methods of engagement are flexible, and the practical and emotional needs of patient partners are addressed throughout the process. Systematic review registration The systematic review protocol was registered with the International Prospective Register of Systematic Reviews PROSPERO (CRD42019127994).


Plain English summary
Patients are experts by experience and are becoming more active as partners on research teams. Patients who are frail and/or seriously ill do not appear to be engaged as research partners to the same extent as those living with more stable illness. The aim of this systematic review was to explore how frail and/or seriously ill patients have been engaged as partners in research.
In 30 studies, frail and/or seriously ill patients were engaged as research partners. They identified: research questions and outcomes important to patients; developed tools and processes more related to patients' needs and experiences; helped collect and/or interpret findings; presented research results; and provided study oversight. Barriers to patients' partnering were mostly related to concerns about their fragile health, their ability to process information and their likely limited ability to partner for the duration of the study due to declining health or death. When frail and/or seriously ill patients were engaged as partners in research, patients had a renewed sense of purpose and felt emotional support, research was more related to patients' needs, and researchers gained greater insight into the lived experience of illness and suffering. Overall, it appears that frail and/ or seriously ill patients can and should be included as research partners. Researchers can work to avoid unduly harming patient partners by being flexible and ensuring patients' physical and emotional needs are addressed during the research process.

Background
Over the past two decades, the commitment to engaging patients as partners in research has steadily gained momentum. International interest in patient engagement has been fostered by the belief that it can enhance the relevance, validity, and quality of research [1]. It is further postulated that research developed in this way will be more applicable to the needs of patients and hence more readily applied [2,3]; thereby, legitimizing research that is often publicly funded [4,5]. Patient engagement has become a moral and ethical imperative and, in some jurisdictions, particularly with marginalized communities, patient engagement also serves as a pre-requisite for research ethics approval [6][7][8]. The engagement of patients as partners in the design, execution and evaluation of health research is now an expectation of several principal funding programs [5,9].
In Canada, as in many other countries, most major national and provincial research funding bodies promote engagement of patients throughout the entire process, from determining the research question to dissemination of the research results [10,11]. A systematic review of 142 studies [2] established that, in most instances, it was possible for patients to contribute their expertise across the continuum of research; however, their engagement tended to be focused in the early stages of the study. The level of patient engagement in the process has varied in intensity and complexity depending on the nature of the research and information needs [11]. A more recent scoping review examining methods and outcomes of patient engagement confirmed that, in the absence of a validated framework, most efforts to engage patients continued to be limited to the early stages of engagement and did not appear to be maintained throughout the lifecycle of most research projects [12].

Engaging patients as research partners
Patients are broadly defined as individuals with personal experience of a health condition [11]. There are numerous terms used for the concept of patient engagement in research including, but not limited to: 'integrated knowledge translation', 'patient and public involvement', 'participation', 'patient engagement', 'public and patient engagement' and 'co-production' [13,14]. Patient engagement can be considered along a continuum from consultation at one end of the spectrum to partnership at the other end of the spectrum of engagement [15]. Research partnership is identified as patient membership on the research team, contributing to shared decision-making across the research process, engaged in the planning, execution and dissemination of research findings [15].
When partnering with patients, there is a shift from the researcher as sole expert to one where researchers and patients are both experts, working together to solve problems and co-generate knowledge [16]. Patients' experiential knowledge (of illness) is not accessible to most researchers, but if leveraged appropriately, has the potential to complement researchers' analytical skills and scientific perspective ( [17] , p. 676). The concept of partnering with patients as equal team members has been demonstrated in clinical guideline development [18], by systematic review teams [19], and in the area of health and services improvement [20]. However, there are ongoing concerns about the need to balance rights to participation with efficiency and outcomes, [21,22], particularly in disciplines that may lack the necessary infrastructure to support patient-facing activities (e.g., preclinical research) [23,24].

Partnering with frail and/or seriously ill patients
Inclusivity is an important principle in meaningful research partnerships with patients and places emphasis on equity of engagement in research [25,26]. Whilst a number of reviews have concentrated on descriptions of the process and methods for various levels of patient engagement, little attention has been directed towards providing detailed accounts of patient characteristics [2,12,15,27]. This oversight makes it difficult to gauge the inclusion of individuals from diverse patient populations.
The vulnerability of certain groups, such as frail and/ or seriously ill patients (e.g., elderly patients with limited functional capacity, patients with high symptom burden, palliative patients), provides additional complexity to the engagement processes for prolonged and more intensive patient partnerships [27][28][29]. Frailty is classified as: a) geriatric condition involving functional decline, with increasing vulnerability to adverse events including mortality, morbidity, disability, hospitalization, and nursing home admission [30], or b) presence of multiple chronic conditions such as arthritis, heart failure, renal failure, and pulmonary disease leading to changes in functional ability [31], or c) presence of cognitive decline and dementia [30,32]. Older adults living with frailty are a diverse group of patients that exhibit physical and/or cognitive impairments. Serious illness is defined as a condition that carries a high risk of mortality, negatively impacts quality of life and daily function, and/or is burdensome in symptoms or treatments [33]. Examples of serious illnesses are cancer (e.g., metastatic or hematologic), advanced liver disease, and advanced pulmonary diseases [33].
Patients who are frail and/or seriously ill have unique needs associated with symptoms related to their condition and/or treatment side effects which may offer researchers' access to a lived experience of illness that is qualitatively different than those with more stable or chronic conditions [29]. Practical issues related to engaging frail and/or seriously ill patients as research partners appears to deter research teams from inviting, or even considering them for membership on the research team [28]. A recent scoping review exploring engagement of geriatric oncology patients found little evidence of patients' inclusion as research partners [29]. Little is known about the engagement of frail and/or seriously ill patients as partners on research teams.

Aim
The aim was to synthesize the evidence on the engagement of frail and/or seriously ill patients as research partners across the research cycle. The specific objectives were to: a) describe the contribution of frail and/or seriously ill patient partners to the stages of the research cycle (and associated research activities), b) identify the barriers and facilitators to partnering encountered by frail and/or seriously ill patients, and researchers, and c) describe the perceived positive and negative impacts of including frail and/or seriously ill patient partners in research from the perspective of patients, researchers, and the research itself.

Study design
A systematic review of qualitative, quantitative, and mixed methods studies was conducted with narrative synthesis. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [34] guided the reporting. The study protocol was developed prior to the literature search and registered via PROSPERO (CRD42019127994).

Guiding conceptual framework
The systematic review was guided by a conceptual framework comprised of two components (see Table 1). The first component utilizes a modified version of the Patient Service User Engagement in Research Framework originating from a prior systematic review by Shippee et al. [15], and addresses patient engagement at different stages of the research cycle and associated activities. The second component addresses the level of engagement in the decision-making process as defined by the International Association of Public Participation (IAP2) Spectrum of Public Participation (see Supplementary file 1 for additional detail) [35]. The IAP2 spectrum denotes five levels of engagement (inform, consult, involve, collaborate, and empower) and has been used in Canada, Australia, New Zealand, Indonesia, Italy, Southern Africa and the USA to outline levels of engagement and promote best practices in patient and public engagement [36].

Data sources and search strategy
An electronic search strategy was developed with the assistance of an experienced health sciences librarian (KF) and adapted for the following databases: MEDLINE® (via Ovid), Cumulative Index to Nursing and Allied Health Literature (CINAHL via EBSCO), Excerpta Medica database (EMBASE® via Ovid), and PsycINFO® (via Ovid). The search strategy included a combination of key words and medical subject headings (MeSH) terms such as "patient engagement", "patient involvement", "patientoriented research". (see Supplementary file 2 for the complete Medline search strategy). Reference lists of the included studies were manually reviewed to maximize the breadth of the review. There were no date limitations. The search strategy was executed from April 4, 2019 to April 6, 2019.

Eligibility criteria
The population, intervention, control, outcomes, study design (PICOS) criteria were used to assess study eligibility [34] (see Table 2). All original studies of any design were eligible if they included frail and/or seriously ill patients as research partners at the level of involvement, collaboration, or empowerment throughout the research  • Frail and/or seriously ill adult patients as per definitions for frailty and serious illness (e.g., elderly patients exhibiting physical and/or cognitive impairments, patients with high symptom burden due to acute illness or treatment effects, acute episodic illness, palliative patients; patients susceptible to adverse events including mortality, morbidity, disability, hospitalization, and nursing home admission).
• Studies where patients were excluded due to frailty of condition (physical and or cognitive) or deemed too ill to participate during acute episodes of serious illness or treatment. • Patients not identified as frail or seriously ill, i.e., survivors, chronic disease (focus on single disease without description of acuity/ severity of condition). • Participants from broader community or public engagement (with no descriptors of frailty and serious illness) • Patients for whom there were no descriptors of physical characteristics or cognitive status. • Pediatric and youth patients (< 18 yrs).

Phenomenon of Interest (I)
• Engagement of frail and/or seriously ill patients as partners in research, i.e., at the level of involvement, collaboration, empowerment.
• Engagement of patients as objects of study, i.e., doing research on or to. • Engagement that took the form of informing patients of research activities, or at the level of consultation only.

No comparator
Outcome (O) • Methods and timing of engagement (i.e., stage(s) of research process). • Level of engagement.
• Engagement strategies, factors associated with barriers and facilitators to engagement. • Positive and/or negative impacts of engagement on patient(s), researcher(s), research and/or ethical concerns.
• Primary research outcomes where patients were research participants only.

Study Type (S)
• Peer-reviewed qualitative, quantitative, or mixed methods studies.
• Studies reported in non-peer reviewed journals.

Language
No language restrictions.
cycle (see Table 1). There were no language restrictions. In order to limit duplication, all systematic reviews were excluded after manually searching the reference lists of relevant reviews. Commentaries and editorials were excluded as well as studies that did not provide any details on patient perspectives or patient condition and when no full text was available.

Study selection
Search results were uploaded to Covidence Systematic Review Software [37]. Following the removal of duplicates, citations were screened independently by two reviewers (CL, JL) based on title and abstract (level 1 screening) and full-text articles (level 2 screening). The studies were assessed against the inclusion and exclusion criteria. Full-texts that did not meet the eligibility criteria were excluded and the rationale was documented in the Covidence Systematic Review Software to facilitate ease of tracking and reporting.

Data extraction
Data extraction forms were developed to provide a standardized and transparent method for examining the methodology and findings from the studies [38]. The forms were piloted on a subset of relevant papers that were included in the review and refined to ensure the extraction template met the specific objectives of the review. The following general characteristics were extracted: year of publication; title, aim, study design, country of conduct; number of frail and/or seriously ill patients engaged in research; patient condition with regard to reports of serious illness and/or frailty of patients. Engagement in research was extracted on four components: a) stages of the research cycle and associated activities within those stages; b) the level of engagement in the decision-making process, i.e., involvement, collaboration, and empowerment (see Table 1); c) barriers and facilitators to engaging frail and/or seriously ill patients in research; d) the described impacts of engaging frail and/or seriously ill patients. Data were extracted by two independent reviewers (CL, JL) and discrepancies resolved through discussion. A third party (DS) was available in the event that consensus could not be reached.

Data analyses
Narrative descriptions were reported for all studies. Data were synthesized in accordance with the guiding conceptual framework, i.e., engagement during the research cycle and by level of engagement. No meta-analyses were conducted as the aim was to identify the scope and types of patient engagement. Additionally, the heterogeneity across studies regarding the design, patient populations, methods, measures used, and a lack of numeric outcomes reported inhibited meta-analyses.

Quality assessment
Two independent reviewers (CL, JL) critically appraised included studies using the updated Mixed Methods Appraisal Tool (MMAT) [39]. The MMAT has been content validated, tested for inter-rater reliability and is increasingly utilized in the quality appraisal of systematic reviews of mixed studies [40][41][42][43]. Scores are based on criteria, which differ according to study type. Each study was appraised according to the criteria met and were ranked as having low, moderate, or high quality. But they were not excluded on the basis of low quality because the overall aim was to identify the scope and types of patient engagement. Reviewers resolved discrepancies through discussion and consensus.

Search and selection results
There were 14,062 citations retrieved from electronic searching (see Fig. 1) [44]. After removing duplicates, 8763 original articles were screened, 431 full text reports were reviewed for eligibility, and 28 studies plus two additional studies identified through manual screening of reference lists in the included studies for a total of 30 studies met eligibility criteria. Included studies were published between 2006 and 2019, with a trend of increasing publications over time; 73% of studies were published within the last 5 years since 2014 (see Fig. 2).

Patient partner research roles: research stages and activities
No studies reported engagement of frail and/or seriously ill patients at the level of empowerment, i.e., the provision of primary direction and governance to a given research endeavor. The highest level of engagement was reported in four of 30 studies (13%) where collaboration     Research prioritization reported in prior publication [74] was demonstrated across all four stages of the research cycle (see Table 4). Patients in these studies partnered in activities including, but not limited to: delineation of the scope of the partnership, contribution to study design, co-leadership on working groups during study execution, data analysis, dissemination activities, and adoption of decision-making roles on research steering/advisory committees [50,52,57,66]. Seven studies (23%) included patients in research priority setting at the broader level of biomedical specialty/ disease/condition, rather than at the individual study level [48,49,54,58,61,67,68]. These studies included patients who were representative of the condition as partners on research steering/advisory committees and who contributed to shared decision-making across the research study cycle.
Eleven studies (37%) described collaboration with frail and/or seriously ill patients across the latter three stages of the research cycle (preparation, execution, and translation) at the individual study level [7, 45, 47, 51, 53, 59, 62-64, 71, 72]. Patients partnered in a variety of different activities that mostly included: assistance with grant applications, input into study design, co-design of project materials, recruitment strategies, data analysis, dissemination activities, and decision-making at research steering/advisory committees. Four (13%) studies included frail and/or seriously ill patient partners from research collaboratives or networks who assisted with the identification of appropriate patient engagement strategies specific to frail and/or seriously ill populations at a broader system level. Patients described contributing to grant writing, proposal development, tool refinement, conducting interviews, representing research findings, and co-authorship on papers across different studies [24,46,55,56].
Four studies (13%) described patient roles during key stages of the research process, rather than across the research cycle [60,65,69,70]. Patients participated in activities at the execution stage of the research cycle, where they piloted research tools, served as peer interviewers, assisted in other forms of data collection, or interpreted data sets [60,65,69,70].

Barriers and facilitators to partnering with patients System level factors
The most commonly cited barrier for researchers to partner in research with frail and/or seriously ill patients was resource constraints, including financial concerns, human resource capacity for support, and the time commitment required for meaningful engagement (15/30) [7, 24, 45-47, 54, 56, 59, 60, 62-64, 69, 71, 72] (see Fig. 3). Researchers also cited lack of formal infrastructure and policy, poorly defined governance mechanisms, and inconsistent processes to support meaningful patient partnerships as a system level barrier (4/30) [24,53,59,64].

Team level factors
Lack of role clarity and expectations related to the contribution of patients throughout the research cycle was cited as a barrier to meaningful engagement by both patients and researchers (6/30) (see Fig. 3) [24,45,53,59,63,69].

Researcher level factors
The most commonly cited perceived barrier of researchers to partnering with frail and/or seriously ill patients in research was related to their concerns about patients' potential lack of continuity in contributions throughout the research cycle due to deterioration in patients' health or cognition, or death (11/30) (see Fig. 3) [7,45,47,53,55,57,59,61,68,70,72]. The second most common barrier was researchers' uncertainty about the value or overall benefit of patient engagement, particularly given the outcomes of the partnership on research may not be visible for some time (9/30) [24,46,52,53,56,57,62,69,72]. Other researcher barriers were perception that research outputs identified by patient partners may not be fully aligned with the initial objectives of the project or might be too costly to implement (4/30) [24,60,63,67], concern for placing additional or perceived unnecessary burden on patients (3/30) [47,52,65], and lack of familiarity and confidence in patient engagement, particularly where patients assume a partnership role (2/30) [56,62].
Perceived negative impacts for patients were cited as potential physical and/or cognitive fatigue related to the effort required during engagement (5/30) [7,47,53,55,72]. Increased emotional vulnerability and the potential for distress in reliving their illness and related negative experiences were also cited as perceived negative impacts to patient as partners (5/30) [7,47,55,71,72].
The negative impacts described by researchers engaging frail and/or seriously ill patients as partners in research were described in relation to the potential strain on scarce resources (particularly related to funding and the human resource capacity required to support patient engagement activities) (15/30) [7, 24, 45-47, 54, 56, 59, 60, 62-64, 69-71]. The additional complexity of the process and time required for engaging patient partners was also cited as a potential impediment to advancing project objectives and meeting timelines closely aligned with research funding cycles (2/30) [7,64].

Study quality
All 30 studies provided evidence of relevant sources of data appropriate for the research question and used a research design relevant to address the research question. Of the 20 qualitative studies (66%) in the review, most were rated as high quality using the MMAT [47,50,52,53,55,57,59,63,66,[69][70][71][72]. Seven (7/20) of the qualitative studies were rated moderately lower because it was difficult to determine whether interpretation of the results was sufficiently substantiated by data [24,45,51,54,56,62,64]. For the two quantitative studies (6.7%) [46,65], there was a risk of nonresponse bias in both studies, particularly in one study where those deemed too ill were excluded from the opportunity to participate [65]. For the eight mixed methods studies (26.7%) [7,48,49,58,60,61,67,68], five had risk of nonresponse bias [48,49,58,61,67] and three reported interpretation of the qualitative results that was not sufficiently substantiated by the data [48,61,68] (see Table 6).

Discussion
The overall aim of this review was to synthesize the evidence on the engagement of frail and/or seriously ill patients as research partners across the research cycle. The 30 studies included in the review provide an indication of an upward trend in the inclusion of frail and/or seriously ill patients as partners in research over the past decade, with a marked increase in the number of studies in the past 5 years. Most studies included patients with cancer, with fewer studies partnered with patients who had dementia and/or frailty, or patients with palliative care needs. There was evidence of research partnerships with frail and/or seriously ill patients across the research cycle. These activities engaged patients on research related to setting priorities, selecting outcomes considered important to patients, grant review, tool development, research conduct, and dissemination of findings. These findings lead to the following three key points for discussion.
There is ongoing deliberation about the paucity of evaluation of patient engagement in research, especially the long-term impacts related to research implementation and ongoing use of research findings [12,22,76,78]. Interestingly, no reports of negative impacts on the research itself were found in the studies included in this review, which may reflect a bias in over-reporting positive impacts of patient engagement, or may suggest that evaluation efforts are more focused on short and intermediate term impacts of partnering with patients [79]. Insufficient evaluation and poor reporting of the negative impacts of patient engagement are described elsewhere in the literature and point to lack of methods and rigorous evaluation tools [22,80]. Without validated evaluation frameworks and consistent identification of both positive and negative outcomes, there is a risk that anecdotal accounts, and perceived barriers to partnership will dominate the discourse of engagement and undermine the successes [79,81].
Given the population of focus, it was surprising to have identified few negative impacts on patients. Negative outcomes were primarily defined as potential depletion of physical and emotional resources, and the likelihood of experiencing emotional distress through reliving painful illness experiences, exposure to undesirable information, or experiencing the direct suffering of others [45,47]. It is difficult to establish whether the limited number of negative impacts identified is due to lack of evaluation or lack of reporting [12,22,28]. However, while a limited number of negative impacts were cited, the possible magnitude of these impacts should not be under-estimated and every effort is required by research teams to mitigate these potentially deleterious impacts. Similarly, when examining the potential impact on researchers partnering with frail and/or seriously ill patients, it is interesting to note that researchers described exposure and sensitization to the lived experience of illness and suffering, yet failed to acknowledge the concomitant emotional labor and associated burden that invariably comes with exposure to suffering [63,82]. Issues of loss and grief are readily acknowledged for patient-partners following a decline in health or the death of others on the team [7]. However, it would appear that feelings of grief and loss, and the subsequent impact to emotional well-being, is not as readily acknowledged for researchers [82]. Failure to address these issues may leave many researchers ill-prepared to deal with emotionally demanding and difficult situations, cause unintended harm, and serve as a deterrent for both patients and researchers alike.
Evaluating the impact of partnering with frail and/or seriously ill patients is essential; limited evidence suggests that patients experienced several positive impacts, particularly when more intensive levels of engagement occurred. The potential emotional benefits described by patient partners (e.g., a renewed sense of purpose whilst coping with a disease over which they have little control, and/or the emotional support from peers on the research team) may in fact serve as a protective factor against emotional distress and vulnerability, and may also serve to quell researchers' hesitation in partnering with them [7,45,47,51,64].

Discontinuity of contribution
Consistent and predictable contribution is an important consideration for teams embarking on a partnership with patients, more so for those involving frail and/or seriously ill patients on research teams. Concern for wellbeing is critical and is cited as a barrier to both initial and ongoing engagement. Discontinuity of contribution is a commonly anticipated barrier to engaging those most frail and/or ill (i.e., patients receiving palliative care, those with progressive dementia, or experiencing aggressive disease progression) [53,70]. Patients' contribution will be lost or interrupted most often due to deterioration in their health or death, and it is incumbent on researchers to mitigate this. Paradoxically, discontinuity of contribution is rarely acknowledged when related to an improvement in condition, and yet, with advances in treatment approaches, particularly within oncology, many serious illnesses beyond the acute treatment phase are now considered chronic conditions [83]. If the purpose of including frail and/or seriously ill patients as partners in research is to provide access to the lived experience of their illness and leverage that knowledge to shape the research that is produced, the concept of discontinuity of contribution needs to be expanded to include situations when patient partners move from serious illness into remission, cure, or survivorship. The transition from serious illness to a period of more stable illness undoubtedly shifts the perspective and lived experience of patients. As such, it may be argued that over time they become less able to speak to the immediate lived experience of serious illness and more acute suffering. As patients are invited to participate in all stages of the research process, it is important to ensure patients within various stages of the illness trajectory are provided with equal opportunity to partner in the very research that is intended to benefit them [84].

Weighing up the costs of partnership
There are moral, ethical, and practical reasons to engage frail and/or seriously ill patients as partners in research [85]; but researchers need to consider whether the impact or benefits of their engagement is warranted by the supplementary costs they will inevitably incur [79]. Facilitators for partnering with frail and/or seriously ill patients will invariably involve additional investments of time, money, and human resources to compensate for the accompanying administrative and emotional burden that research teams undertake in the endeavor [2,28,77]. Appropriate funding must be made available to teams dedicated to engaging frail and/or seriously ill patients as research partners, particularly when factoring in the need to address patients' emotional and physical needs throughout the course of engagement [7,47]. Therefore, it is necessary to optimize efforts at patient engagement to ensure expertise of patients who truly represent illness across the trajectory, particularly with regard to frailty and/or serious illness.

Strengths and limitations
The diversity of nomenclature describing patient engagement combined with a deficiency of standardized reporting and lack of specific indexing may have resulted in some relevant studies being undetected [2,15]. There are distinctions between what constitutes a "patient," "service user," or member of the "public," which pose additional methodological challenges for identification, recruitment and reporting [86]. Moreover, trajectories of disease progression, acute episodic exacerbation, and aggressive treatment regimens create challenges for defining frail and/or seriously ill patients [87]. To mitigate the challenges generated by issues of nomenclature and the potential fluidity of patients' condition, the search strategy was designed intentionally to be broad in order to cast a wide net for potentially relevant papers.
Further effort was taken to review the reference lists of the included studies and recently published reviews on patient engagement. To mitigate potential bias two independent reviewers were involved during study screening, data extraction, and critical appraisal. The reviewers met numerous times throughout the review process to discuss and remain consistent. All supporting files were reviewed, attention was paid to descriptors of patient condition, and associated published study protocols, where available, were traced and reviewed. Of particular relevance, one of the reviewers was a patient who was representative of being seriously ill, experiencing illness and high treatment burden at the time of the review. The second reviewer works in a direct clinical role with frail and vulnerable populations. Co-authors have clinical expertise in oncology, palliative care, frail elderly care, integrated knowledge translation, systematic review methods, and community based participatory research and were instrumental in further addressing clinical and methodological issues during the review. There were few studies reporting on the engagement of frail and/or seriously ill patients. Hence, in the spirit of transparency and inclusion, none of the lower quality studies were excluded. Interestingly, quality issues in the quantitative studies were related to the potential for non-response bias whereby those deemed too ill were not engaged as research partners.

Conclusion
Engaging frail and/or seriously ill patients as research partners has offered research teams a unique insight into understanding what it is like to live with a debilitating and fragile condition to develop research that more accurately addresses their needs. This review provides limited, but promising evidence that it is possible to successfully engage frail and/or seriously ill patients as partners in research without causing them harm. However, researchers need to ensure the purpose of engagement is well-defined, the timing and methods of inclusion are flexible, and the practical and emotional needs of patient partners are addressed. This review also highlights the need for more rigorous reporting of patient characteristics alongside the experiences, benefits, harms and impacts of their engagement in order to build best practices for engaging this vulnerable population.