Selecting interventions for a psychosocial support program for prostate cancer patients undergoing active surveillance: A modified Delphi study

Abstract Objective Curative treatment of low‐risk prostate cancer (LR‐PCa) does not improve cancer specific survival and active surveillance (AS) is recommended. Although AS is cost‐effective and reduces treatment‐related complications, it requires psychosocial support. Research on psychosocial interventions specifically focused on men undergoing AS is limited. Aim of this study is to reach consensus amongst relevant stakeholders on selecting interventions offering psychosocial support to PCa patients during AS. Methods In accordance with the RAND/UCLA method, a modified Delphi approach was used to establish consensus on selecting interventions. During phase one, interventions were identified through a literature review and open survey among all participants. During phase two, three consensus rounds were conducted to rate potential interventions and obtain statistical consensus. The IQ healthcare consensus tool was used to calculate statistical consensus. Results After the first consensus round, 31 participants scored individual interventions on relevance using a 9‐point Likert scale resulting in the selection of six interventions. During the second consensus round 13 discussion items were reviewed during a focus group. After the third consensus round, seven additional interventions were selected by 23 participants. Conclusions In total, 13 interventions were selected for inclusion in a support program. This included four interventions within the domain information and education, three within coping and support, one intervention within physical wellbeing and four within the domain lifestyle.


| Design
A modified Delphi approach is used to establish consensus on interventions selected for inclusion in a support program. In contrast to the more exploratory character of classical Delphi methods, interventions in this study were previously identified through literature review and stakeholder consultation. 18,19 This study was conducted according to the RAND/UCLA Appropriateness Method. 20

| Participants
A stakeholder analysis identified relevant stakeholders (e.g., patients, nurses, paramedics, physicians and mental health professionals).
Purposive snowball sampling allowed inclusion of participants representing all stakeholders from various parts of the Netherlands.
Potential participants were informed about the study and approached via e-mail or telephone. Informed consent was obtained.
At least five stakeholders per subgroup were enrolled to ensure stability of results. 21

| Phase 1: Selection of interventions
Between August and October 2020, a literature review was conducted. 15 In addition to this review, relevant interventions described in a report from the Dutch Society of Psychosocial Oncology (NVPO) were added to the intervention list. 22

| Phase 2: Consensus rounds
During consensus round one, participants received the intervention list through LimeSurvey. Participants scored all interventions on relevance using a 9-point Likert scale. Relevance in this study was defined as the extent to which participants found interventions essential for a support program. An open text field was added to each intervention and enabled participants to place remarks or suggestions. Participants were invited to complete the rating process between 9 July and 20 September 2021.
The second consensus round took place on 12 January 2022 and consisted of a 2 h focus group meeting. Due to COVID-19 restrictions the meeting was organized online using Microsoft Teams.
In accordance with the RAND/UCLA method the meeting was led by an independent experienced moderator (MN) and at least nine panel members were invited to participate. 20 Using purposive sampling, a proportional representation of all stakeholders was ensured. Prior to the meeting, all focus group participants received a list of discussion items. After obtaining permission from participants the meeting was recorded. To emphasize the importance of a patient-centered approach, each discussion item was introduced by the moderator after which patient perspectives were explicitly asked. Afterward, all other participants were asked to comment on the discussion item and more timid participants were actively invited to present their point of view. A researcher (KD) was present as an observant, taking field notes, writing down memo's and meeting minutes.
All study participants were briefed on the focus group results.
During consensus round three, participants received a modified list of interventions through LimeSurvey. Again, all participants were invited to score interventions on relevance using a 9-point Likert scale. An open text field was available for additional remarks. Participants were invited to complete the rating process between 3 March and 24 March 2022.

| Data analysis
The IQ healthcare consensus tool was used to calculate statistical consensus after the first and third round. 23 In accordance with the RAND/UCLA method, participants were invited to rate each intervention on relevance using a 9-point Likert scale. Scores ranged from 1 not relevant to 9 very relevant. The RAND/UCLA method provides clear guidelines on execution of a consensus study and its procedures have been refined since its development in the 1980s resulting in a reliable and rigorous approach. 23 The IQ healthcare consensus tool is based on the RAND/UCLA method and uses median and highest tertile scores.
After analyzing data from round one, a selection, discussion or no selection label was appointed to each intervention (see Table 1). A subgroup analysis was conducted to calculate consensus scores for the specific subgroups and identify between-group differences. In addition, remarks and suggestions placed in open text fields were collected, and thematically analyzed. This provided insight into the rating process, existing concerns and motives. Results from the dataanalysis were discussed during a research team meeting. Interventions with a discussion or no selection label, substantial between-group differences and important remarks and suggestions were selected by the research team as discussion item for the focus group meeting. Important results and findings from round one were presented to all study participants via e-mail.
During the second round, discussion items were presented to panel members. Interventions were adjusted if focus group participants deemed it necessary or otherwise removed. The moderator checked for consensus on adjustment or removal. Consensus was established if >75% of participants agreed on adjustment or removal.
During the focus group meeting, notes were collected by a researcher (KD) and discussed within the research team. Meeting notes and suggested adjustments were returned to all focus group participants for a member check. After approval, the list of interventions was modified in accordance with suggestions and findings from the first two rounds. After collection of data during the third consensus round, data were analyzed and a final consensus label was appointed to each intervention: selection, discussion or no selection.

| Incomplete data handling
To prevent loss of valuable data, both complete and incomplete surveys were included in the analysis.

| Selection of interventions
A total of 39 interventions were selected for the first consensus round (see Supporting Information S1). Interventions were divided into four domains (see Table 1). A literature review identified 13 interventions, 12 interventions were derived from the NVPO report and 14 resulted from the open survey amongst experts.

| Participants
A total of 43 subjects were invited to participate in this study. This included 11 patients, nine nurses/nurse practitioners, 10 urologists, five paramedical and eight mental health professionals. Participation was declined by four subjects (one patient, two urologists and one nurse) due to practical reasons: time, personal circumstances.
Approximately 80% (31/39) of subjects returned the survey. Three patients did not complete the entire survey. The primary reason provided was their self-described inability to rate specific interventions.

| First consensus round
After analyzing data from the first consensus round, statistical consensus on selection, no selection or discussion on all 39 interventions was established. Consensus on selection of six interventions was reached (see Table 1).

| Thematic analysis of open text fields
Analysis of open text fields demonstrated three recurring themes: intervention feasibility, suggestions for adaption and tailored care.
Feasibility concerns were often expressed in regard to sexological interventions. Especially costs and reimbursement possibilities were questioned. In addition, availability of certified sexologists was a major concern. Q1: "Costs! This is not reimbursed by health insurance companies and in a lot of hospitals sexologists are not present." Time was also often addressed as a limiting factor. For instance in regard to multidisciplinary consultations.  The relationship between anxiety, uncertainty, and exercise was questioned by some participants. The importance of food and diet as a self-management strategy to cope with the psychosocial burden of AS was recognized by some participants and doubted by others.
Presence of sexological or urological problems in the AS population was not acknowledged by some participants. They linked these problems to patients in active treatment groups.

| Second consensus round
The second consensus round consisted of an online focus group meeting. Out of the 12 invited focus group members, 11 participated. This included three patients, two nurse practitioners, two urologists, three paramedical caregivers and one mental health professional.
One mental health professional was unable to attend. During the meeting, all 13 predetermined items, were discussed. Based on group consensus, four interventions were removed and nine interventions were adjusted (see Table 2).

| Third consensus round
Based on results from the previous two rounds, an adapted list containing 14 interventions was returned to study participants (see Table 3). Out of the 39 participants, three declined participation due to practical reasons. The rating process was completed by 23/36 participants through LimeSurvey resulting in a 64% response. This included five patients, four nurse practitioners, five urologists, three paramedics, and five mental health professionals. After data-analysis, statistical consensus on all 14 interventions was established. All 23 participants completed the survey. Consensus on additional selection of seven interventions was reached (see Table 3). After the third consensus round, seven remaining interventions were rejected based on their discussion or no selection label. Due to the small subgroup sizes a subgroup analysis was not performed.

| DISCUSSION
Aim of this study was to reach consensus amongst relevant stakeholders on selecting interventions for a psychosocial support program offered to PCa patients undergoing AS. Findings from this study suggest that this support program should include 13 interventions.
After round one, immediate consensus was reached on application of decision aids, SDM and provision of monitoring protocols. This demonstrates the widely shared desire to improve information and education during treatment decision making in LR-PCa. The need for transparent, unambiguous, tailored information and education involving both patients, partners and family members has been underlined in previous literature. 9,24 This is becoming more important since AS is increasingly applied and also extends to patients with favorable intermediate risk PCa. 25 However, merely implementing informational and educational interventions is not enough to actively involve and empower patients. Research suggests patient-centered decision making also requires a behavior change in caregivers. 26