Examining the effectiveness and implementation of patient treatment decision‐aid tools for men with localised prostate cancer: A systematic review

Men diagnosed with localised prostate cancer (LPC) often face a difficult process deciding on a treatment choice that suits their personal preferences. This systematic review examines the impact of patient treatment decision‐aids (DAs) on decisional outcomes and treatment choice for men diagnosed with LPC. Our secondary aim was to examine how DAs have been implemented into routine clinical practice.


| INTRODUCTION
Prostate cancer (PC) is one of the most commonly occurring cancers on a global scale, where it was estimated there were up to 1.3 million diagnosed cases reported in 2018 alone. 1 Men diagnosed with localised PC often find it difficult to make a treatment decision as they have a variety of treatment options to choose from where there is no clear best choice. 2,3 Instead, this decision is based on the patient's own personal preferences and lifestyle. Specifically, potentially curative treatments for localised prostate cancer (LPC) are highly invasive and can lead to debilitating and long-term side-effects such as urinary incontinence and erectile dysfunction. 4,5 In contrast, active surveillance (AS) can also be offered to men as a management strategy for their PC where their tumor is closely monitored, while delaying radical treatments. 2,6 Patients who undertake AS have been shown to have equivalent survival outcomes compared to those receiving invasive treatments. 7 However, some men who consider AS have also reported feeling anxious about not receiving curative treatment and showed a long-term decrease in health-related quality of life. 2,8 Consequently, decision-related distress is common among men with localised PC who need to deliberate on the potential side effects of invasive treatments while also considering the risk their cancer might worsen through AS. 9 To address these issues, treatment decision aids (DAs) for LPC have been developed. As defined by the International Patient Decision Aids Standards (IPDAS) Collaboration, treatment DAs are tools that provide evidence-based information to assist patients in making a treatment decision, while also taking into consideration the patient's personal values and preferences. 10 These tools can be presented in a variety of formats, including textual, audio, visual or web-based. Decision aids can vary in the decision support they provide, but generally they clearly indicate to the patient that a treatment decision must be made, including having no treatment; provide evidence-based information concerning the diagnosis, treatment options, and associated benefits and harms; either implicitly or explicitly inform patients that their decision is sensitive to their personal preferences and values; and help clarify what the patients' preferences and values are when weighing up a treatment choice. [10][11][12] To date, there have been a handful of systematic reviews focusing on the effectiveness of PC treatment patient DAs on the decision-making process. These systematic reviews varied in findings with one review suggesting DAs were capable of improving knowledge and actively involve patients in the decision-making process, 13 while two other reviews indicated DAs showed mixed results with decisional outcomes. 3,14 However, these systematic reviews used a broad definition of a DA, where they did not distinguish between DAs that were created specifically for patient use and interventions that required input or participation from research or clinical staff such as interview consultations and multidisciplinary clinics. Additionally, none of these reviews assessed or reported on the process and/or effectiveness of outcomes associated with the implementation of the DA within the study setting. We argue that these reviews examined PC treatment decision interventions that do not fit the internationally accepted definition of DA tools 15 and are further limited as they provided no contextual summaries or evidence on how implementation of the DA was facilitated. These omissions deemphasize the role patients' values and preferences play in decision making, thus limiting the interpretation and practical knowledge of the systematic DA findings.
The aim of this systematic review is to summarise and critically evaluate the literature regarding the effectiveness of treatment DA tools, as defined by internationally accepted DA criteria for men with LPC on key outcomes of decision-making and treatment choice. 15 The secondary aim was to detail reported implementation outcomes related to the integration of the DA in clinical practice.

| Data sources and study selection
The systematic review followed the guidelines of Preferred Reporting Items for Systematic Reviews and Meta-Analysis 2020 guidelines. 16 Inclusion criteria were publications that: 1) focused on evaluating the effectiveness of DAs as an intervention on decisional outcomes or treatment choice; 2) involved men newly diagnosed with localised lowrisk (i.e., non-aggressive cancer; Gleason score 6) and/or intermediate risk PC (i.e., moderately aggressive cancer; Gleason score 7) 17 who had to make a treatment decision; 3) investigated DAs which were paperbased (e.g. hardcopy booklet), online-based (e.g. Internet website), computer-based (e.g. CD-ROM) or video-based (e.g. DVD) and 4) were original publications in English language. Studies that included different risk groups of PC had to report the results of the risk groups separately to be included in the review. Exclusion criteria were: 1) literature reviews, books, unpublished articles, theses, commentaries, protocols, and abstracts of conferences, or 2) studies focusing on DAs for men with advanced PC, prostate-specific antigen screening, or 3) publications that described treatment decision interventions delivered by clinical or research staff, as well as multidisciplinary clinics, that is, interventions that were not specific decision support tools as defined by IPDAS. 15 As eliciting data on implementation outcomes was a secondary aim, studies were not excluded if they contained no data on implementation outcomes. and Web of Science. The starting point of our search strategy was to include broad terms relating to decision making and DAs to encompass papers that assessed various decisional outcomes. We did not want to limit the search to specific decision outcomes in case there were newer decision outcomes that had not yet been assessed in previous systematic reviews. Search terms used were ("prostat* cancer" OR 470 -TODIO ET AL. "prostatic neoplasm*") AND ("decision* aid*" OR "decision* tool*" OR "decision* support*" OR "decision* instrument*" OR "decision* technolog*" OR "decision* program*" OR "decision* intervention*") AND ("patient participation" OR "patient involvement" OR "decision making" OR "choice behavior"). Full search strategies for each database are summarised in Appendices (see Supplementary File).

| Data extraction, quality assessment and synthesis
After removing duplicates and excluding clearly non-relevant articles based on titles and abstracts, ET retrieved the remaining potentially relevant full-text articles. These were read in full and assessed for eligibility using the Covidence software program. 18 Any uncertainty regarding eligibility was resolved through review and discussion between two reviewers, ET and JS.
The Mixed-Method Appraisal Tool (MMAT) 19 was used by ET and CM to assess methodological quality of selected publications.
Two screening questions and five assessment criteria were applied to each study, which gave a score ranging from 0% (no criteria met) to 100% (all criteria met). Discrepancies were discussed among reviewers when scores differed. The two reviewers independently assessed and rated the studies, where final scores were discussed until agreement was reached on most (>80%) of the appraisal items.
The first author (ET) identified and extracted information from the studies that related to key decisional outcomes and treatment choice associated with DAs for localised PC using Covidence. Data on how decisional outcomes were assessed and affected by DAs at different time points was synthesised from each study. These outcomes were reviewed and discussed among the authors until the final set of decisional outcomes were agreed upon. The review strategy was approved and registered with PROSPERO (ID: CRD42021243634).
Data regarding implementation outcomes was extracted in accordance with the taxonomy of implementation outcomes described by Proctor et al 2011. 20 Relevant data were aligned with the eight conceptually distinct implementation outcomes of acceptability, adoption, appropriateness, feasibility, fidelity, implementation cost, penetration, and sustainability. If an eligible article mentioned an additional or supplementary paper that discussed implementation outcomes, those references were followed and assessed/extracted to ensure completeness of our reporting. Methods used to measure implementation outcomes and key findings were extracted by CM & AM. Where available any mentions of specific implementation strategies were noted.

| Study selection
The searches identified 1742 records (Medline = 215; Embase = 156; PsycINFO = 140; CINAHL = 239, Cochrane = 1, Scopus = 614, and Web of Science = 377). Three additional articles were manually identified through review of previously published reviews and were included as they met the eligibility criteria for this study. After removal of duplicates, 851 titles and abstracts were screened, of which 729 were identified as nonrelevant and excluded.
In total, 122 full-text articles were assessed for eligibility. From these, 98 articles were excluded for not meeting the eligibility criteria.
A total of 24 articles 21-44 were included for the analysis (7 nonrandomised studies, 16 randomised control trial (RCT) studies, and one qualitative study). Three articles 21,23,26 had their data combined with trials reporting results from the same study 22,24,27 leaving 21 original studies. The PRISMA flowchart is presented in Figure 1.

| Study quality
Each publication was assigned a quality score using the MMAT criteria. 19 Quality scores of the quantitative studies ranged from 40% to 100%. Typically, the lower scores in the RCT studies stemmed from poor reporting of methods, particularly with randomisation and blinding procedures. Lower scores in non-RCT studies were largely due to confounders in design and analysis not being accounting for.
One RCT 31 study and one qualitative study 35 included in our analysis scored 100%.

| Study characteristics
The selected studies were published between 1998 and 2022. Table 1 summarises the general characteristics of the included studies.
Thirteen were conducted in North America, seven in Europe, and one in Asia. The recruitment settings varied with most studies using a combination of hospitals and urology/oncology clinics. Three studies recruited patients only from urology clinics. 28,32,44 Various studies specifically recruited from a multidisciplinary genitourinary oncology clinic, 38 a telephone-based cancer information service, 30 and veterans' administration centers. 31 One study did not specify the recruitment setting. 34 A total of 5296 men participated in the studies included in this review. The mean age of participants ranged from 62 to 71. Patient demographics revealed patients tended to be well-educated, where most participants had some college education or graduated college (range, 56%-70%), and varied greatly in terms of patients who were retired or still employed (range, 24%-80%). Most patients reported they were married/partnered (range, 54%-98%) and were white or Caucasian (range, 59%-90%), apart from one study conducted in Malaysia, 36 where 50% of patients were of Chinese descent and another study that focused on minority men where 54% of men were Black/African American. 44 One trial included a subset of high-risk patients, 37 but for this review only the low-and intermediate-risk Usual care for the control group varied among RCT studies where several trials specified that they provided men readily available TODIO ET AL. educational resources and links to reputable websites, [21][22][23][24]37 while some studies stated that usual care was standard consultation visits 36,38,44 or decisional counseling as per the center's normal practice. 26,27,29,40,43

| Decision-aid characteristics
The DAs differed in presentation and/or format (see Table 1). Most DAs were designed as a computer/web-based multimedia program, while three studies tested a DA booklet, 25,31,36 and two other studies presented the DA as a video/DVD format. 33,41 One study 34 measured a plain language DA in three mediums: booklet, audiotape, and Internet. Most DAs discussed treatment options including "watchful waiting" (treating symptoms as they arise) or AS (planned, regular surveillance of the cancer), surgery options (open/robotic-assisted radical prostatectomy), and radiation therapy (external beam radiation, brachytherapy). Four studies also included the treatment option of hormonal therapy. 31,32,41,42 Sixteen studies used a control intervention to assess the DA intervention. Thirteen studies included usual care as the control group. Davison and colleagues 28 compared an individualised DA with a generic DA, while Taylor and colleagues 42 assessed an interactive CD-ROM DA with an information only DA. One study compared a simple booklet DA to a complex booklet DA. 31 Holmes-Rovner and colleagues 34 used historical controls to compare their DA intervention. A recent study used a cluster RCT design to compare two different online/multimedia DAs before and during their clinician visit, as well as having a usual care arm. 44 Decision aids were typically aimed at educating and coaching patients on their PC diagnosis, the different treatment options available, and encouraging shared decision-making with their clinician. Five studies mentioned their DA included an explicit values clarification exercise to help patients weigh-up their personal preference for a specific treatment. 26,27,32,36,39,40 Five DAs were customisable or tailored programs that showed individualised information that fit the patient's diagnosis or preferences. 21,22,[26][27][28][29]41 Three DA interventions were tools that assessed the patient's preferences to generate an individualised information report that helped recommend a treatment option to be discussed with their clinician. 37,38,43

| Effect of decision-aids on decisional outcomes
Six key decisional outcomes were identified and examined in the 21 selected studies (see Table 2): diagnosis and treatment knowledge, F I G U R E 1 Preferred reporting items of systematic reviews and meta-analysis (Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA-P)) study selection flowchart.  41 One study focused on assessing patients' knowledge regarding the rationale for AS using two multiple choice questions, 33 and patients who viewed the DA were 2.9 times more likely to better understand the rationale for AS than those who did not view the DA.

| Preparation and involvement in decisionmaking
Several studies examined whether their DA intervention was able to help better prepare patients to make a decision about their treatment choice, as measured through the Preparation for Decision Making scale. 45 Berry and colleagues 21 highlighted that the use of the DA intervention did not significantly predict higher perceived preparation for decision making, but instead preparation was associated with obtaining resources prior to being involved in the study and being from an ethnic minority. Furthermore, Jalil et al. (2020) 36 found both the intervention and control groups were more prepared to make a treatment decision post-intervention, but no statistical differences were found between groups. The authors of these studies 21,36 suggested this could be due to limitations in their methodology, where patients perhaps already felt prepared to make a decision through their consultation with their clinician as well as their use of other information resources prior to the use of the DA. In contrast, another study found that patients in the DA intervention were significantly less prepared to make a decision compared to the control group, with this finding attributed to patients having high expectations of the DA. 26 The patient's perceived involvement in the decision-making process has also been assessed with two studies using studyspecific questionnaires 32,35 while three studies 28 pre-and post-trial conditions, 32 with patients overall preferring to play an active role in their treatment decision. Furthermore, a single arm study showed most patients felt they participated "a lot" in their treatment decision after accessing the DA, 41 while the qualitative interview study 35 revealed patients showed more active involvement in the decision-making process and made a more deliberative choice. Results from one RCT indicated that patients who used the DA were 79% more likely to prefer making an independent treatment decision rather than shared decision making with their clinician compared to those who did not receive the DA. 42 In contrast, a different RCT study found that patients exposed to a simple DA booklet showed greater interest in shared decision-making before their clinician consultation compared to patients receiving a complex DA booklet. 31

| Treatment choice
Decision aids across studies appeared to have conflicting effects on patient's final treatment choice. Four older RCT studies 22,28,42,43 revealed no significant differences in treatment decision between intervention groups, and a non-RCT sudy 39 indicated 67% of patients did not change their treatment preference after the use of the DA.
One RCT found that patients were more likely to consider conservative treatment when provided a simple DA booklet versus a complex DA booklet before their clinician consultation but found no changes in treatment received in the 6-month follow-up. 31 It is important to note that these older studies did not include AS (i.e. planned active monitoring of disease progression) as an option in their DA or included the outdated term "watchful waiting" which  47 and two older studies using a study-specific questionnaire. 28,41 Two studies that used the SWD 25,37 reported patients in the DA group had higher SWD scores compared to the control group and these scores remained significant 6 months to 2 years after the treatment decision was made. 37 Onel and colleagues 41 reported similar results, where 93% of patients surveyed were highly satisfied with their chosen treatment after viewing the DA. In contrast, Berry and colleagues 21 found SWD was not related to DA use 6 months later using multivariate analysis, but was instead predicted by information sources used before baseline and higher perceived preparation at 1 month. Davison and colleagues 28 also found no difference in satisfaction with treatment choice between patients using a generic or individualised DA after 4 weeks.

| Decisional conflict
Decisional conflict was the most common patient outcome measured in the included studies. Six studies 22,25,26,30,38,43 used the original Decisional Conflict Scale (DCS), 48 while five studies 23,28,36,42 employed a short form version of the DCS. Diefenbach and colleagues 29 used a study-specific scale adapted from their previous study. 49 Overall, five studies indicated patients who received the DA showed decreased decisional conflict compared to the control group. 23,25,28,29,38,43 However, the heterogeneity of the timing in these studies cannot be ignored; with two measuring decisional conflict after 1 month, 23,36 one measuring after 3 months, 25 and three studies assessing immediately after accessing the DA. 29,38,43 Five other studies found no statistically significant differences in total DCS scores for either groups over time ranging from 1 week to 6 months. 22,26,28,30,42  all showed consensus that decision regret did not differ between intervention and control groups. 21,24,27 Instead, factors such as anxious or depressive symptoms, 27 education, marital status, preparation for decision-making at 1 month, and bowel function at 6 months 21,24,27 were significant predictors of decisional regret.
However, one study that used the regret subscale of the Memorial Anxiety Scale for PC 51 revealed patients in the intervention group showed a greater decline in decisional regret than the control group after 24 months. 37 3.5.7 | Implementation outcomes Data on at least one implementation outcome were available and extracted in 11 of the included studies ( Table 3). The most commonly reported outcomes were acceptability (n = 7) and appropriateness (n = 7). No data on implementation cost, penetration, and sustainability were available or discussed by any included article. Additionally, information as to an explicit implementation strategy (i.e. a detailed plan on how to facilitate the integration of the DA within the patient clinical pathway) used for DA dissemination and uptake was not reported by any study.
Methods to collect implementation outcomes often included surveys (n = 8), qualitative assessments (n = 1), other (n = 1). "Other" pertained to a metric tracking software embedded in the DA. In all included studies implementation outcomes were only assessed with patients.
Three studies had a published protocol paper. The relevant protocol papers were examined to determine if outcomes related to implementation were planned for assessment. Of the three protocol papers, one explicitly detailed plans to collect data on the "optimal implementation" of the DA. 52 These included an evaluation of information provision (appropriateness) and opinions on the DA (acceptability). However, our search did not find published results from this protocol.

| DISCUSSION
The purpose of this review was to examine the effects of patient -487 in treatment choice with DA use. Therefore, DAs involving a values clarification exercise component may be effective for patients with low-risk PC who have to make a more critical decision on whether to choose between deferred treatment or curative treatment; however, further investigation is needed.
The relationship between DA use and the decisional outcomes of treatment satisfaction and decisional conflict also remains inconclusive due to heterogeneity of the studies. With regards to treatment satisfaction, studies used different statistical analyses where one study used multiple regression and found no relation with DA use and treatment satisfaction, 21 while other studies that used t-tests found differences in satisfaction after DA use. Therefore, it may be important for future studies to use more complex analyses to investigate the relationship with DA use and satisfaction and how this is related to other factors such as social support. 21 Although most of the studies used the same instrument to assess decisional conflict, the time differences in following up patients varied greatly between studies which made it difficult to summarise this outcome by time-periods. Furthermore, heightened decisional conflict may be evoked regardless of using a DA due to the effortful decision-making process patients must go through with considering all treatment options and weighing up the pros and cons to suit their personal preferences and values. 26,57 Consequently, it is disputed whether existing DAs decrease decisional conflict. 58,59 DAs appeared to have no effect on decisional regret across most studies. Although DAs can educate and coach patients to make a treatment decision, it seems that other factors such as health outcomes play a stronger role in influencing regret. 21,24,27 Around half of the included studies assessed at least one implementation outcome. The majority of these discussed either acceptability or appropriateness of the DA tool. Penetration, sustainability, and implementation costs were not reported on and to our knowledge (according to available study protocols) not planned for assessment. It appears the literature on DAs for PC is largely still focused on design and assessment of effectiveness. This is accordance with Proctor et al's 20 idea of salience by implementation stage which denotes that certain implementation outcomes are assessed in accordance with the timeliness of development of the evidencebased intervention. 20 For example, acceptability and appropriateness of an intervention is typically assessed during design and innovation, while sustainability is considered at mid-to late stages of implementation. However, despite certain implementation outcomes better lending themselves for assessment at later stages of implementation journeys, it has been recommended that scoping for all implementation outcomes can be considered at any stages of innovation development, typically referred to as "designing for implementation". 60 Assessing key stakeholder considerations on what would be required to make a DA part of routine practice, for example, would be a light touch sustainability insight that may aid in future development of implementation strategies for DAs. Furthermore, healthcare provider perceptions of implementation of DAs were not assessed by any included study. Acknowledging that DAs are typically a resource that healthcare professionals typically provide to patients, it is important to understand provider perspectives of acceptability, feasibility, and fidelity to best place the DA within clinical practice.

| Study limitations
The publications analyzed in this review had several limitations.
Most notable is the variations in the instruments used to assess outcomes. As Lin and colleagues 13 point out, there are still no agreed upon standardised instruments to assess outcomes after DA use. Therefore, it would be beneficial for future research to identify the best outcome measures to assess DA use and perform these in a consistent manner to yield more robust evidence.
Furthermore, RCT recruitment settings varied with the types of "usual care" provided. Consequently, it is likely that some patients received more substantial care compared to other control groups that were recruited from more general clinical or community settings.
A limitation of this review is that the publication date in the eligibility criteria was not limited, where older trials included outdated therapeutic options, as previously mentioned. Thus, it is recommended that future reviews consider focusing on papers that include the most current contemporary therapeutic options available in their DAs to improve homogeneity across studies. Moreover, we understand that our systematic search may have missed some papers that focused on a particular outcome as our search did not include key decisional outcome terms, for example, "regret". We suggest for future systematic reviews evaluating decisional outcomes of PC DAs to keep this in mind when developing their search strategy. Further, the quality of the information contained in the DAs themselves was not assessed in this systematic review, this was because DA content was not publicly available. Thus we could not verify whether the DAs met the quality standards established by the IPDAS Collaboration. 15

| Clinical implications
Overall, this review indicates DAs may improve patient knowledge but have no significant impact on decisional regret or preparedness in decision-making. Furthermore, there were mixed findings among the different DAs in relation to decision-making involvement, decisional conflict, treatment choice, and satisfaction. Thus, it appears that DAs are not fulfilling the expectations of PC patients in most aspects of the treatment decision-making process. Nevertheless, more current DAs have shown the potential to influence low-risk PC patients to choose conservative management over curative treatments. Future research could adopt an inductive qualitative approach using indepth interviews to gain a greater insight into how patients perceive DAs, particularly with how they feel DAs should be structured and the type of information they want to be included in a DA.
This would facilitate a holistic understanding of men's experiences 488 -TODIO ET AL.
using DAs and may identify themes that could be explored and measured as quantitative decisional outcomes that have not already been assessed in previous research. Refining the design and quality of DA research would also inform the implementation of DAs in clinical practice, where based on the studies analyzed, very few studies provided in-depth analyses on how their DA could be implemented in clinical practice. Alternatively, other outcomes may be more appropriate to examine in future reviews such as health outcomes and quality of life, which could provide different insights into the benefits of DAs other than the various decisional outcomes that have previously been examined.
Our findings suggest that much improvement is needed on study design to test effectiveness of DAs before implementing in practice. It is recommended that future studies adopt an effectiveness-implementation hybrid design as this would allow for DAs to be assessed on both outcomes to ensure that the DA intervention is beneficial in routine practice moving forward. 61 This review also highlights the importance of ensuring that DA content is regularly updated to keep up with current clinical standards that are constantly evolving with newer acceptable treatments and practices.
The shortcomings associated with studies examining PC treatment DAs on decision-making have been highlighted. Evidence indicates most DAs appear to improve patient knowledge but fail to show robust benefits relating to decisional regret and decisionmaking preparedness. Furthermore, contradictory findings relating to different aspects of the decision-making process, including treatment choice, decision-making involvement, decisional conflict, and satisfaction were apparent. This inconsistency does not necessarily indicate DAs are inefficient, rather it can mainly be attributed to methodological issues among studies, particularly with the variability of non-standardised instruments utilised and potentially the implementation strategies used which may have led to these varying results. Given the difficult and stressful experience men go through with choosing a treatment option, it is paramount that men are provided the resources necessary to make an optimal value-based decision. Thus, it is recommended that future research utilise standardised methods to measure the effectiveness and implementation of DAs appropriately and adequately, as well as explore patients' experiences through qualitative approaches to establish the advantages of using DAs.

ACKNOWLEDGMENTS
Open access publishing facilitated by University of New South Wales, as part of the Wiley -University of New South Wales agreement via the Council of Australian University Librarians.

CONFLICT OF INTEREST
The Authors declare that there is no conflict of interest.

DATA AVAILABILITY STATEMENT
Data sharing is not applicable to this article as no new data were created or analyzed in this study.

ETHICS STATEMENT
This study required no ethical approval as data are publicly available.

PATIENT CONSENT STATEMENT
This study did not require patient consent as only publicly available data were used.