Use of a Delphi process to formulate a pragmatic framework for care planning in older adults with advanced bladder cancer

Abstract

Purpose

New advanced bladder diagnoses usually occur in the 8th or 9th decade of life. Most data on bladder cancer care has been garnered from a younger population. Older adults with cancer are more likely to have comorbid conditions, undetected vulnerabilities, and priorities that value function in comparison to younger adults with cancer. However, functional assessments and value elicitation are not consistently employed at the outset of care planning in this unique population. We proposed the use of an intervention bundle, or framework, to guide early care planning for older adults with advanced bladder cancer to address the delivery of goal-concordant care.

Methods

We followed a modified Delphi process to determine the components of the framework. We undertook an initial literature review and then convened a transdisciplinary group of professionals with internationally recognized expertise in urologic oncology (medical and surgical), clinical trial design, geriatric oncology, palliative care, exercise physiology, or cancer rehab. This panel participated in surveys and meetings to determine representative measures for the (1) functional assessment, (2) chemotherapy toxicity estimate, and (3) guided goals of care discussion components of the framework.

Results

The group recommended the short physical performance battery to measure functional status, the Cancer and Aging Research Group chemotherapy toxicity calculator to estimate chemotherapy risks, and the Serious Illness Conversation Guide to elicit patient values and priorities.

Conclusion

This consensus building process utilized evidence-based measures backed by expert consensus to contribute to a pragmatic framework to guide care planning in older adults with advanced bladder cancer.

Introduction

Older patients who have been diagnosed with a new malignancy will spend multiple visits to review the results of testing that inform discussions on a potential plan of care [1]. The oncologist synthesizes pathologic, radiographic, and health-history data to determine the anti-cancer options available to a patient. However, older adults are a unique population, with increased risks of aging-related vulnerabilities that may influence treatment tolerance and are not generally detected in standard oncologic evaluations [2, 3].

Older adults with bladder cancer tend to have more comorbid conditions, altered organ function, and goals that prioritize preservation of independence. Standard oncology evaluation of functional status misses nuanced needs of an older patient [2]. For example, the oncology team may not learn that their elderly patient depends on neighbors to deliver meals, or a child to set out medications—signifying underlying frailty that may influence cancer treatment tolerance and management. Additionally, the cancer team relies on trial data to estimate the potential toxicity risks and benefits of a treatment regimen—yet trial populations are younger and healthier than the average older adult oncology patient [3, 4]. Most oncologic care is structured around these limited trial data, and reactive when complications occur, rather than proactively considering the values and needs of this unique patient group from the outset.

There are existing measures and tools to support a proactive approach in caring for older adults with cancer, such as a geriatric assessment and goals of care discussions. There are data showing that a geriatric assessment improves communication, treatment tolerance, and reduces hospital use [5, 6]. A geriatric assessment screens for vulnerability or frailty in domains of health such as cognition, nutrition, physical function, and medication use. Similarly, there are data, primarily in the palliative care literature, that support the early engagement in care planning discussions, prior to end-of-life, for patients to share their values and priorities in the setting of a serious illness. These elements of care are generally not incorporated into early visits with the oncologist [7–9].

These early visits to the oncologists are heavily influenced by the practice patterns and communication skills of the treatment team. We learned of patient preferences when we engaged a group of older adults with bladder cancer in order to learn of their experiences, priorities, and perspectives which have shaped our design of a care planning framework for bladder cancer [10]. In response to the identified variations in care, we conducted a modified Delphi process with clinical experts to assist in the development of a care planning framework that includes the essential items of care planning for older adults with advanced cancer. These clinical experts provided input on the use of evidence-based geriatric assessments and how they approach goals of care discussions on a regular basis for older adults with advanced cancer. This standardized and systematic approach toward identifying best practices is a first step toward ensuring that all patients receive geriatric assessments to inform their care plans, much in the same way they receive accurate staging, pathologic diagnosis, and oncologic treatment education.

The purpose of our modified Delphi process was to provide an evidence-based recommendation on the components of the care planning framework for older adults with advanced bladder cancer. In addition, this framework, guided by pragmatic considerations in initial care planning visits with older adults facing an advanced bladder cancer diagnosis, will ultimately form the intervention of a pilot study of the care planning framework in urologic oncology clinics. Several components to enhance care planning for older adults with a newly diagnosed incurable cancer that may potentially be treated with systemic chemotherapy were included (1) measured physical function assessment, enhancing the standard oncologic screening with geriatric principles; (2) a chemotherapy toxicity calculation to estimate the risk of grade 3 or higher toxicity on chemotherapy; and (3) a planned visit with an advanced practice provider in which patients shared their priorities and life values in order to contribute to a goal-concordant care plan. In this manuscript, we describe the process of finalizing these three components for an intervention package that would be tested in an initial consultation for older adults with advanced bladder cancer (Fig. 1).

Fig. 1.

Example use of the integrated framework in up-front care planning with older adults

Methods

We aimed to finalize the (1) functional assessment, (2) chemotherapy toxicity estimate, and (3) goals of care discussion guide that would be included in an up-front care planning framework. We employed a modified Delphi methodology to decide on the 3 measures to include in the framework with the a priori decision that all 3 measures would be included. This method is a formal group consensus process which systematically and quantitatively combines systematic literature review of evidence and expert opinion by asking content expert participants to rate, discuss, and identify their chosen components of a care planning framework by building consensus [11–14].

The team undertook the Delphi Survey preparation in late 2020 into the first quarter of 2021. During this time, we undertook a literature search, panel identification, and survey development. As mentioned, we had developed the pillars of the ABC123 and utilized this process to identify existing measures to incorporate into the framework. First, we performed a literature review to identify standard measurements of the domains of interest. Additionally, we referenced existing Society of Geriatric Oncology and National Comprehensive Cancer Network guidelines regarding functional assessment, chemotherapy toxicity calculation, and palliative care discussions [15, 16]. The functional measures were chosen from a review of measures in the NCCN and SIOG guidelines as well as a PubMed^® search performed in Fall 2020 and allowing for articles published within 10 years. The following terms were used: Functional change on cancer treatment, in the elderly, “during treatment”: elderly, older adults, geriatric AND walk speed, gait, balance, chair rise, mobility, frailty, fatigue, physical strength, physical function, physical fitness AND “during treatment” or “undergoing treatment” or “on treatment” or “active treatment” or “being treated” or “during chemotherapy” or “undergoing chemotherapy” or “on chemotherapy”. This yielded 39 results which were then evaluated to determine the functional measures listed. We included the top 10 cited measures that could be applied in a clinical setting. The chemotherapy toxicity calculators were identified from 2 prospectively validated choices available and applicable to the study population. These are referenced in the NCCN and SIOG guidelines and were thus used as choices. The authors are not aware of other prospectively validated calculators predicting for graded toxicity. The guided goals of care discussion frameworks were more complex and involved a search of the literature using keywords and phrases including “goals of care,” “guided goals of care,” and “goals of care discussions,” we searched PubMed and located 9 framework models, of which we decided to include 8 for Delphi panel review. We reviewed roughly 150 articles and selected this subset to meet the following criteria: (1) validated clinical tools for provider-patient/family discussions about goals of care and (2) frameworks or approaches that are reasonable to implement and require minimal additional resources or training. The only framework that was located and not included was a model that involved a card game to guide goals of care discussions, which would require special resources in clinical care that are beyond the scope of this work as we aim to develop a model of ABC care that is generalizable to many settings. Geriatric assessment measures, covering all domains, were not included as this pilot framework focused on 3 core tenets rather than the multiple domains that may be screened for or assessed in a geriatric assessment.

We identified up to 10 representative choices of items to be included in the care planning framework guided by the following categories: (1) physical function measures, (2) chemotherapy toxicity estimates, (3) guides for goals of care discussions. These categories were used to align assessment of the older adult’s physical function, current or future potential chemotherapy toxicity to inform a person-centered, values-based goals of care discussion. Next, we followed a modified Delphi method by presenting these standardized measures from the literature review to a panel of experts in geriatric medicine, palliative care, palliative care advanced practice, urologic oncology surgery and medical oncology [10, 17]. The Delphi participants were chosen based on national reputation and expertise in their field, with some clear overlapping commitment to bladder cancer, or older adults, or patient-centered research. The study team predefined a need for expertise in geriatrics, palliative care, physical function assessment in older adults, and bladder cancer oncology. Additionally, we sought out expertise in secondary expertise of advanced practice provider skills, geriatric oncology, and patient-centered research. A representative expert in each role was identified by the author group and approached for participation, or a recommendation of an alternate participant. The authors did draw upon their academic network, yet also worked to include professionals from varied academic centers and from diverse disciplines. This work largely took place in the midst of the COVID pandemic, a time in which clinicians were heavily engaged in clinical care—thus, the panel remained small in an effort to maximize engagement in the pilot design.

The Delphi measures were presented via individual descriptions of the measure, or embedded within a clinical vignette of older adults with advanced cancer. In contrast to the quantitative measures of the assessment of physical function and chemotherapy toxicity, which have reported psychometric properties, the choice of goals of care frameworks was guided by expert opinion and consensus recommendations. We took themes found in the Serious Illness Conversation Guide©, the VitalTalk^® REMAP guide, the House model, the 3 phase model and the Dunlay general guide and highlighted key themes in a clinical vignette [17–21].

To achieve consensus, we asked the Delphi panel to rank the physical function measures, chemotherapy toxicity calculators, and the components of a goals of care discussion (Table 1). These rankings were collected from the panel experts via a Qualtrics online survey tool (see Appendix). The two rounds of anonymous surveys were sent to panel experts and open until 100% completion. Each participant received at least 2 and up to 6 reminders. Items were ranked on a 9-point scale in which a score of 1 signified a measure was “extremely unlikely” to be recommended for inclusion in the care planning framework, 5 as neither likely nor unlikely, and 9 as extremely likely to be recommended for inclusion in the care planning framework. Panelists reviewed and ranked each item and had the opportunity to offer modifications or reasoning behind their decision. Panelists were also able to generate additional items for inclusion in future rounds. The second round of surveys opened with a reminder of the goal of the exercise and respondents were offered the round 1 scores for each survey item next to that choice. No comments were shared with the respondents. Round 2 rankings were again scored and a score of 7–9 signaled agreement for rankings and that the item may be acceptable as content in the care planning framework.

Table 1.

Delphi panel participant characteristics Total N = 6

Area of professional focus
Geriatrics	2
Exercise physiology	1
Urologic oncology	1
Medical oncology	1
Palliative care	3
Geriatric oncology	1
Clinical research	6
Clinical care	5
Degree
MD	4
PhD	3
ANP	1
MS	2
MPH	1
Years in practice
10–15	2
> 15	4

After these two survey rounds, we held a meeting to build consensus and form the final recommendations. A virtual video conference was held to discuss the results, identify measures that the panel most supported for inclusion, and determine if consensus could be achieved to create the 3-item care planning framework. Two study team members moderated the conference to monitor potential bias in the discussion and ensure that all participants were able to share opinions. The sessions began with a reminder of the purpose of this work, contributing to a care planning framework that would be pilot tested for use in patients with newly diagnosed advanced bladder cancer. The study team provided the results of previous rankings, including aggregated summary statistics and evidence-based literature review, to each panel member. In these meetings, the research team prompted members of the Delphi panel to review each survey section—physical function, chemotherapy toxicity, and goals of care discussions. We offered background on the options included in the survey, shared survey data, and allowed the participants to question or comment on rankings. The discussion of each topic required an 80% agreement level amongst participants in order to meet the consensus requirement. The Colorado Multiple Institutional Review Board approved this research as exempt.

Results

The areas of expertise of the Delphi panel are listed in Table 1, and the survey choices are listed in Table 2. We describe the panel feedback on each item with the framework: (1) physical function, (2) chemotherapy toxicity, and (3) guided goals of care discussions. Rounds 1 and 2 of the survey contained the same options; and participants did not suggest any additional options for round 2. Due to scheduling conflicts among the Delphi panel experts, two separate virtual conferences were held with the same study team members were present at both. The summary of data and representative quotations from the consensus meetings are shared below.

Table 2.

Survey choices for Delphi rounds

Area of focus	Measure of choice
Physical function	6 minute walk23,24
	Timed walk23,25
	10 meter walk23,25
	Short Physical Performance Battery28, 29
	30 second sit to stand26
	5 time sit to stand27
	Hand grip30, 31
	Leg press32, 33
Chemotherapy toxicity estimate calculators	Cancer and Aging Research Group Calculator34
	Chemotherapy risk assessment scale for high age patients1
Goals of care discussion frameworks	Serious Illness Conversation Guide18
	REMAP20
	3 Phase Model22
	Dunlay guide21
	House Model19

The physical function measures listed in Fig. 2 were ranked by all participants and included: 6-minute walk test [23, 24], timed walk test [23, 25], 10-meter walk test [23, 25], 30-second sit-to-stand [26], 5 time sit-to-stand [27], short physical performance battery (SPPB) [28, 29], hand grip [30, 31], and single maximum leg press [32, 33]. Based on our defined acceptability for inclusion and exclusion, the 10-meter walk and leg press were below acceptable scores while the 30-second sit-to-stand was above the inclusion cutoff. The SPPB and 5 time sit-to-stand were scored just at the acceptability cutoff (Fig. 2).

Fig. 2.

Physical function evaluations of rounds 1 and 2 of the Delphi survey. Median scores with minimum and maximum

TUG: Timed up and go; 30s: 30 seconds; Min: Minute; SPPB: Short Physical Performance Battery

Each measure was reviewed during the consensus discussion. The discussion of physical function assessment focused on the practicality of the assessment as well as the potential for garnering highly predictive information from a brief and clinically practical single assessment. These are some examples of discussion points that aligned with the rankings of these measures:

I was thinking well what is my pragmatically most likely to yield the best information in the shortest amount of time, and I might actually in clinic…so 30 seconds sit to stand that seemed pretty short, so it was a favorite vote

what we’re really trying to do is unearth previously undiscovered frailty

For me … it’s efficiency and it’s need not to have extra equipment

The group favored the sit-to-stand and walk tests as measures that could be performed in the clinic and were well validated. After a discussion to achieve consensus, the group agreed on the SPPB as it included both the sit-to-stand and walking tests and is performed without equipment or the need for dedicated time, space, or training of the administrator.

The Cancer and Aging Research Group (CARG) chemotherapy toxicity calculator [34] and the chemotherapy risk assessment scale for high age patients (CRASH) [35] were presented as general measures and then demonstrated in clinical vignettes. The tools were consistently ranked lower than an acceptable range for inclusion (Fig. 3). Given the a priori decision to include a toxicity estimate in the framework, the group then came to consensus on the preferred calculator for inclusion. The group agreed the Cancer and Aging Research Group calculator was more easily implemented in a clinical setting. More people favored the CARG over the CRASH. Below are examples of discussion points that aligned with the rankings of these measures:

…the CARG has more practical questions to me

I’m not convinced that toxicity is the appropriate outcome…and maybe I don’t care as much about tools to predict the outcomes

Fig. 3.

Mean chemotherapy toxicity calculator scores in rounds 1 and 2 of the Delphi survey, with minimum and maximum scores displayed

The panel responded to vignettes containing themes found in the Serious Illness Conversation Guide©, the VitalTalk^® REMAP guide, the House model, the 3 phase model, and the Dunlay general guide [18–22]. These themes were consistently ranked highly (Fig. 4). We did offer a framework which included visual aids, and the scored responses excluded this option.

Fig. 4.

Evaluations of components necessary for goals of care discussions in rounds 1 and 2 of the Delphi survey (median scores with minimum and maximum displayed)

In the consensus meeting, the group reviewed each example of the goals of care discussion guide in more detail and even discussed either creating a “de novo” guide versus using an existing resource. The conversation also centered upon the timing of the framework implementation and the planned usage at the outset of treatment planning, in comparison to when care plans may need to change. Below are examples of discussion points for these goals of care discussion guides:

It’s remarkable how consistent these responses stayed from round to round, compared to the others

The… model is not practical

It seems that best evidence base is behind the serious illness conversation guide and Re-MAP

If you follow the guide, you are most likely to hit all of the elements

Based on the review of each goal of care discussion guide, the group consensus was to move forward with the Serious Illness Conversation Guide© as it covered the highly ranked theme and seemed most applicable to conversations occurring at the outset of care planning—the planned timing of the care planning framework use.

Discussion

Through a modified Delphi process, we identified key measures for use in an initial oncology care planning framework specific to older adults with advanced bladder cancer. The final evidence-based, expert-backed recommendations favored the short physical performance battery (SPPB), CARG chemotherapy toxicity calculator, and the Serious Illness Conversation Guide© as representative measures of function, toxicity risk, and guided goals of care discussion.

Given the nature of the modified Delphi design, the overall purpose was to leverage the expertise of panelists to identify the content for this care planning framework. Our approach was further strengthened by the inclusion of transdisciplinary authorities. The panel was not limited to medical oncologists or geriatricians. The panel included the perspective of multi-specialty physicians, advanced nurse care, and exercise physiology. Each panelist had experience in direct care of older adults with cancer, as well as expertise in clinical research, and offered feedback which leaned on their experience and specialty. The diverse perspectives of the panel offered rich discussions and yielded a pragmatic framework balancing reliable measures with practicality of delivery in variable clinical settings. For example, while the panel did favor a timed walk test or sit-to-stand test, they noted that the short physical performance battery included both measures; and thus, may be more informative in the long run without adding too much burden to the clinical team. Similarly, the Serious Illness Conversation Guide © offered a robust guide for a goals of care discussion that could be followed by generalists, rather than specialists.

Throughout the Delphi survey rounds, we collected general comments from respondents. Of interest was the round of comments concerning the chemotherapy toxicity calculation tools. These are validated tools to predict the likelihood of an older person experiencing a high-grade adverse effect due to cytotoxic chemotherapy. Many oncologists use clinical trial data as a reference for adverse events of an anticancer regimen. However, these studies generally exclude older adults. The group did not see as much value in the use of these calculators, either arguing that Common Terminology Criteria for Adverse Events (CTCAE) Grade 3 chemotherapy toxicity may not be an outcome of importance to older patients or that the data from published trials could be references or that the other domains (such as function) may outweigh this calculation. It was also noted that this calculator applies only to cytotoxic chemotherapy and excludes estimates of immunotherapy and targeted therapeutics. The group perspective was that these calculators may offer information to inform shared decision discussions rather than informing a prescribing pattern. These comments may inform future work using these calculators. In contrast, the comments and discussion around the Goals of Care themes were also notable for generalized support. The vignettes offered were meant to highlight either themes of a goals of care discussion or nuanced clinical scenarios which may alter some of the core components of a goals of care discussion. Despite these changes, the group continued to rank each theme highly citing the applicability to all patients with advanced cancer.

This process integrated independently beneficial measures into a single framework may inform care planning at the outset and lead to higher quality care. Much like cancer staging or pathologic data, this “basic measure” has the potential to be implemented in various practice environments and offer objective data to patients and their cancer care teams. Through integration into a common framework, this has the potential to limit the contribution of clinical bias and limit implementation variables. While these measures are not novel, the inclusion into a single framework that can be referenced in routine oncologic care is highly innovative. Additionally, these measures are not bladder cancer specific and have the potential to apply to additional populations of people with advanced cancer. This work was then evaluated in a pilot study (NCT05408507) with results pending [36].

Our work had multiple limitations including the panel makeup based on expertise, yet with only one representative per clinical field. For example, we included one medical oncologist, one geriatrician, and one exercise physiologist, etc. The oncology clinicians were also focused on the care of patients with bladder cancer; therefore, it is unclear how these recommendations may apply to other tumor types. Additionally, the study was designed for a group meeting consensus discussion, yet it was difficult to coordinate all schedules, and so we held two separate consensus meetings. In each meeting, the expert panel engaged in an in-depth discussion. However, the second meeting was a combination of discussion and “fact checking” of the first group’s recommendations. Third, the physical function measures and chemotherapy toxicity calculators were highly cited and validated, but the guided goals of care discussions had less well-defined measurement properties. Much of the existing literature is built around advanced care planning or when clinical status and care plans are shifting—with less guidance as to learning of patients’ values at the outset of care—which limited our ability to draw from guides used in similar situations.

Conclusion

This consensus building process has resulted in a pragmatic framework, incorporating evidence-based and expert-backed recommendations to guide care planning in older adults with advanced bladder cancer. Next steps involve conducting a pilot study to test the feasibility and acceptability of utilizing this framework to inform care planning and the impact on goal-concordant care in older adults with newly diagnosed advanced bladder cancer.

Supplementary Material

The online version contains supplementary material available at https://doi.org/10.1007/s00520-026-10532-1.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.