Decision Aids for Decision Making about Locally Advance Breast Cancer: A Systematic Review

Abstract Locally advanced breast cancer (LABC) is a subset of breast cancer with locoregional progression without distant metastasis. The multimodality treatment (surgery, chemotherapy, radiotherapy, hormonal and targeted therapy if required) could significantly improve results in this specific group of patients. The complex and multiple options of treatment with similar mortality rates but different outcomes depending on the patient’s desires, preferences and social environment require aid to facilitate the individual patient’s decisions (e.g. Decision Aids (DAs) targeting patients considering primary or adjuvant treatment in LABC). In this context, DAs have been proven fundamental to help patients and clinicians share and agree on the best value option. The current systematic review aimed to evaluate the existing DAs related to these patients with LABC and identify current status and possible improvement areas (possible scarcity and heterogeneity of instruments, the status of their development, explanation of their purpose,…). No previous systematic reviews have been published on this topic. Following Prospero registration no: CRD42021286173, studies about LABC DAs were identified, without data or language restrictions, through a systematic search of bibliographic databases in December 2021. Quality was assessed using Qualsyst criteria (range 0.0–1.0). The quality of the 17 selected studies ranged from 0.46 to 0.95. Of them, 14/17 (82%) were DAs about treatment, only one (6%) about diagnosis, and 2/17 (12%) about the employment of DAs. No screening or follow-up DAs were retrieved. Twelve (70.6%) DAs were online tools. They varied broadly regarding their characteristics and purposes. Most of the studies focused on developing and testing different DAs (5/17; 29.4%) and their impact (7/17; 41.2%). Only 4/17 (23.5%) analysed their implementation and cost. These instruments have proven to improve patient’s knowledge and decision-making, decrease patient anxiety, and patients tend to undergo treatment. However, nowadays, there is still a need for further research and consensus on methodology to develop practical DAs.


Introduction
Breast cancer (BC) is the most common and deadly cancer in women (1,2). However, its morbimortality has declined in current years due to earlier diagnoses and new therapies and strategies (3,4). Locally advanced breast cancer (LABC) is a heterogeneous group that includes the most advanced breast tumours in the absence of distant metastasis. It includes bigger than 5 cm size tumours with regional lymphadenopathy (N1-3), any size tumours with chest wall or skin invasion, regardless of regional lymphadenopathy and the presence of clinically fixed or matted axillary lymph nodes or any of infraclavicular, supraclavicular, or internal mammary lymphadenopathy regardless of tumour stage (5). The clinical treatment of locally advanced breast cancer is complex. Nowadays, incorporating biomarker data (histologic grade, molecular subtypes, and multigene assays) into the classic anatomic tumour, axillary nodes, and metastasis (TNM) staging could satisfactorily inform clinical management of this heterogeneous disease. These diverse molecular subtypes (luminal, triple negative or HER2 positive) generate specific targeted therapy for each patient and anticipate the response and prognosis of patients treated with preoperative therapy (6). Patients with HER2-positive or triple-negative disease had improved pathological complete response rates compared with the luminal subtype. However, they were related to particularly poorer survival, especially in those with residual disease after preoperative therapy. The standard treatment for LABC has usually been preoperative chemotherapy followed by reassessment for trying conservative surgery (7)(8)(9). However, due to the enormous heterogeneity of LABC, different treatments, including neoadjuvant chemotherapy, surgery, and radiation therapy, have been used against this type of cancer with comparable overall survival but differences in recurrence risks and cosmetic outcomes (10,11). Therapeutic decision-making should be tailored to the individual patient and tumour's characteristics (type, subtype, size, location, … ), considering patients' preferences and beliefs (4,6,12). Different treatments have been used against this type of cancer with comparable overall survival but differences in recurrence risks and cosmetic outcomes (10,13). Therapeutic decision-making should be individualised, considering patients' preferences and beliefs (4,12).
In this context, shared decision making (SDM) is crucial as it is an essential high-quality cancer care pillar. It becomes essential when there are different combinations of treatment alternatives with overall equivalent potential (e.g. neoadjuvant þ lumpectomy vs mastectomy) that can produce diverse outcomes depending on patients' preferences and values (14,15). It becomes especially required when various treatment alternatives with overall equivalent potential can render diverse results depending on patients' preferences and values (14,15). Its execution in clinical practice has continuous barriers (11,(16)(17)(18) and remains poor (19). Decision aids (DAs) are tools or interventions to promote SDM, helping to elucidate individual preferences and values when they would be crucial for the decision choice (11,14,20). They could be an analogic or digital instrument, system, technology or interactive decision support. DAs in clinical practice enhance people's understanding of the risks and benefits of a treatment and may help them make choices in line with their personal values and preferences (11).
Based on a literature search, an evaluation of DAs for LABC treatment choices has not been previously reported. Our systematic review evaluated the existing DAs related to patients with LABC, their characteristics and usefulness, and identified possible improvement areas.

Methods
A prospective protocol was designed for this systematic review based on recommended methods for literature searches and registered on the Prospero register of systematic reviews (CRD42021286173). Our systematic review was conceived under the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement (PRISMA) (21) (see Appendix S1).

Data sources and searches
A systematic search of databases (Scopus, Trip database, Web of Science, EMBASE, CDSR) was conducted without languages or date restrictions in December 2021. It incorporated Mesh terms "shared decision making," "aid decision," "decision support tools," "breast cancer," "locally advanced cancer," and including word variants (see Appendix S2). We have also explored the list of references of related publications to recognise potential extra and suitable studies to collect.

Study selection and data extraction
We selected original studies, both randomised clinical trials (RCT) and observational studies, that reported data related to the patient characteristics, use and impact of a patient diagnosis and treatment DA for LABC.
We excluded clinical practice guidelines, consensus statements, protocols, reviews and meta-analyses, case reports, editorials, comments, and conference abstracts. We also rejected publications focused on early-stage and metastatic breast cancer due to disparities in the treatment purpose of the decisions or articles that did not specify the stage. There were no limitations on genre, histological (high-low grade) and molecular type of cancer (luminal, basal-like or HER2-enriched).
After removing duplicate publications, the eligibility of each title and abstract was independently scrutinised by three reviewers (RPTA, LM and ARH). Any controversies or inconsistencies were resolved by discussion between reviewers and a fourth reviewer (MMC) that helped reach a consensus. After this step, they selected the studies that met the selection criteria for a fulltext assessment working independently.

Data extraction and synthesis
Two authors (MMC and RPTA) recorded essential information using a piloted data extraction form concerning characteristics of studies: year of publication, publication in a journal, quartile of the journal, entity related to the study, design/method, level of evidence according to Sackett,Straus and Richardson (22), if it is a study about LABC only, area of study, the aim of the study, characteristics of the decision support, DA type, DA related area, comparator of the study, population, number of participants (n), outcome (s)/results and key findings and comments. A descriptive analysis of quality assessment items concerning quantitative and qualitative studies was performed. Quality was appraised using the Qualsyst score system (23).

Quality assessment and risk of bias
The Qualsyst scoring system was used to evaluate the quality of the included studies (23). This tool was used for the appraisal of methodology quality in a variety of study designs, so it implemented two scoring systems: for evaluating quantitative and qualitative research reports. Quantitative and qualitative studies assessed 14 and 10 items, respectively (Appendix S3) and scored from 0 to 2 depending on the degree to which the specific criteria/item were met (i.e., "yes" ¼ 2, "partial" ¼ 1, "no" ¼ 0). Appendix S3 reported the checklist of items studied. In quantitative papers, if the item was not applicable to a specific study design was marked as "n/a" and was excluded from the summary score calculation. A summary score was calculated for each publication by summing the total score obtained across items and dividing by the maximum possible score; being the formula: (total number of possible itemsnon applicable items) Ã 2. We employed the definition of the quality of a study using the Qualsyst score defined by Lee et al. (24): strong (Qualsyst score >0.80), good (Qualsyst score 0.71-0.79), adequate (Qualsyst score 0.50-0.70) and limited (Qualsyst score <0.50). Three reviewers (RPTA, ARH and LM) independently evaluated the quality of studies using a piloted data extraction form. An Inter-rater agreement coefficient (ICC) was calculated to study the reliability of the reviewers, considering excellent ICC > 90% (25). After this calculation, disagreements between the three authors over the risk of bias in studies were solved by a group discussion that tried to reach a consensus. If consensus was not reached, an arbitrator (MMC) was involved and took the final decision. This final unified result was used to calculate the ultimate quality score. Publications were also ranked by the level of evidence suggested by Sackett, Straus and Richardson (Appendix S4) (22).

Study selection
This systematic research found 1535 citations from online databases; we removed 36 duplicates and 1448 publications for not meeting the selection criteria. Fifty-one studies were included after filtering through reviews of titles and abstracts. Utterly, after the full-text assessment, a total of 17 studies (26-42) (1 qualitative and 16 quantitative) were selected to include in our systematic review ( Figure 1) while 34 were excluded for not accomplishing the selection criteria (11 guidance documents, 9 protocols, 3 systematic reviews, 1 editorial, 3 comments, 4 conference abstracts and 3 studies about metastatic breast cancer). Of these sixteen quantitative studies, we found 6/17 (35.3%) descriptive studies and 10/17 (58.8%) clinical trials (3 non-randomised clinical trials (NRCT) and 7 RCT). Table 1 shows the studies' characteristics (author and year, area of study, DA type, type of BC process, population, number of participants, aim of study, outcome and score). Only one research was specifically LABC related (1/17; 6%) (33). See Appendix S5. Regarding the area of study, 14/17 (82%) were focused on DA treatment-related, only one (6%) was about a DA for diagnosis, and finally, 2/17 (12%) were about the employment of DAs. Most of the included articles (15/17; 88%) were from 2010 onwards.
The majority of the studies (10/17; 58.5%) showed that DAs improve knowledge and reduce decision conflict (29)(30)(31)(32)(34)(35)(36)(38)(39)(40)(41). DAs could be used as informational and therapeutic tools that facilitate self-reflection and self-healing (31). Table 2 describes the characteristics of DAs that appeared in the selected studies. Six DAs with a specific name specified were reported in eight manuscripts, while six works did not specify a proper name for the DA reported. Three studies did not focus on any particular DA (37-39), so they did not appear in the table. Although 12/17 DAs (70.6%) were online tools, their characteristics varied broadly. Most of the tools (9/17; 52.9%) were focused on the treatment (2/17 general, 4/17 surgical or 3/17 adjuvant treatment). No DAs related to screening or follow-up were found.
Lee et al. (33) developed a general DA (an online comic) to help doctor-patient dialogue through a BC patient's diagnosis and treatment process. Belkora et al. (38) suggested the necessity of a practitioner's more active role to ensure that patients receive adequate help according to their specific situation, their preferences and values. Raphael et al. (26) highlighted that the use of DA by patients could depend on the format these DA were presented, so if a multidisciplinary team report stated that treatment "had to be discussed with the patient," patients would be more likely to use the DA. The patient's literacy was not related to the probability of using a DA.
The acceptance of the use of the DAs was challenging to study due to the heterogeneity of the studies selected. However, it was good (range 76%28 to 85%42) in the two publications that studied it. Volz et al. (37) and Belkora et al. (38). confirmed that when the participants reviewed some or all of the DAs provided in each study, patients were keener to develop different proposals for the decision-making process. Some of them (81%) made a list of questions, recorded the clinical consultation (20%), and took notes (66%). One of the main advantages obtained with the use of a DA was that the decision-making process was more bearable and easier versus in usual care processes, where the information   obtained was described as "overwhelming and difficult to process" (32,36). The influence of different factors on the use of DAs was studied too. Regarding demographic factors, the high-educational level was related to a better understanding and could modify the decision choices (41,42). Using a native language in the tool was crucial to better adhesion, use, and understanding of the tool (36 (35) studied BRECONDA, a tool that provided information concerning reconstruction options and supported the decision-making process allowing the user to self-assess and clarify her values regarding this decision. It noted that the use of a reconstruction DA for helping in the decision conflict could be cost-effective. Politi et al. (28) developed the BREASTChoice for surgical treatment decision conflict concerning reconstruction regarding booklets format and also studied cost information's influence on the patient's primary treatment decision with two developed tools that provided or did not provide cost information (27). It demonstrated that patients wanted treatment cost information, and DAs should help them understand it. Institution policy should guide clinicians in improving the quality of these cost discussions.
Regarding Adjuvant treatment, Belkora et al.  (42) studied Adjuvant!, an online tool that delivered recurrence and mortality risk predictions for patients with BC considering adjuvant therapies based on patient and tumour characteristics. They obtained a similar result, but regarding adjuvant treatment. Only 58% of women in the DA group chose adjuvant therapy, compared to 87% of their usual care counterparts (p < .01). DA demonstrated a more efficient selection of treatment, particularly for patients for whom adjuvant treatment has little benefit (29,42).

Quality assessment
The quality appraisal by Qualsyst score (23) for the qualitative study was 0.70, while quantitative studies ranged from 0.42 to 0.95; (possible range 0.0-1.0). Table 3 specifies Qualsyst items for every selected study. The qualitative study (33) had an adequate quality (Qualsyst score 0.7). Regarding quantitative studies, 9/16 (56.3%) were classified as strong (Qualsyst score >0.8), 5/16 (31.3%) were reported as adequate (Qualsyst score 0.5-0.69), and one (6.3%) as good (Qualsyst score 0.7-0.79), and another also as limited (Qualsyst score <0.5). Although RCTs should hold the highest scores, in our systematic review, RCTs reported similar quality (range 0.68-0.93) to descriptive studies (range 0.67-0.95). Studies that achieved lower on Qualsyst items usually did not define whether and how researchers and participants were blinded; they reported outcomes poorly, did not use validated development measures, and did not report a clear justification.
The ICC for quantitative studies was 99% (R1-R2 ¼ 0.99; R1-R3 ¼ 0.99; R2-R3 ¼ 0.99). There was only one qualitative study, so we thought ICC would not be useful. Disagreements occurred in item 2 "appropriation of the study design" with two reviewers stating "partial" and one "yes", and item 5 "sampling strategy described and justified" with two reviewers scoring "yes" and one "partial".
In the qualitative study (33), the main aim, the clarity of the context, the sampling strategy and data collection of the study, the connection to a theoretical framework or broader body of knowledge and the conclusions supported by the results were notorious, while the reflexivity of the account was not clearly or systematically described.
In quantitative studies, the results and conclusions were supported in sufficient detail in all the studies. Methods, comparison groups, outcomes,    (30) and Yao et al. (29) were the best-qualified studies with 0.95, respectively.

Main findings
Our systematic review found that LABC DAs were very heterogeneous and demonstrated that they were essential for high-quality patient-centred cancer care as they enhanced knowledge and reduced patients' decision conflict. The purpose of the selected studies was varied, and they were all from 2010 ahead. The majority of the studies focused on describing the type of DAs and the difficulties of their use than their impact or patient satisfaction. Most of the DAs were related to treatment, and there was only one study about a DA specifically related to LABC. The quality assessment in qualitative and quantitative studies was varied but mostly adequate or better (Qualsyst score >0.7). More than a third of the DAs created were online tools, and their characteristics also differed considerably.

Strengths and weaknesses
One of the most important strengths of our work is that we have done an exhaustive systematic review focused on general LABC management (diagnosis, treatment and follow-up) with no language or data source limitations. A prior study (43) about a different but related topic (DAs in early BC treatment) was done several years ago, retrieving articles from 2011 to 2015.
One perceived limitation could be the heterogeneity of the selected publications, which makes the comparison between studies difficult. However, this heterogeneity itself may be unavoidable if we want to obtain a broad perspective of the current situation of DAs on LABC, as studies differ in purposes and methods.
The subjective character of data extraction involving the quality score of the studies could be another presumed limitation of this review. We minimised this problem by using three reviewers who explored the assessment tool manuals to develop a mutual knowledge of the scoring methods and triplicating data extraction. A fourth independent arbitrator helped reach a consensus where there were disagreements. The Qualsyst tool was suitable for estimating quality, but it was not a proper instrument to determine specific study biases. This could result in overestimated studies' quality scores. RCTs should hold the highest Qualsyst scores, primarily due to more prominent study design methodological rigour, sampling process, and grander detail in reporting results, including variance estimations, but in our systematic review, descriptive studies reported similar quality. We have followed directions regulated by Qualsyst authors and previous studies to minimise errors. A prior validated cut-off (24,44) used and published in previous research was employed to classify the studies' quality.

Implications
SDM is essential for high-quality patient-centred cancer care (15,(45)(46)(47)(48). BC management decisions could be challenging for both clinicians and patients, and there is a need for individualising cancer care, taking into account management options and patients' preferences and values. Diverse approaches may offer a priori equivalent advantages and disadvantages, whereas possible outcomes are deeply related to the patient's values and personal situation (45,49). Although SDM application in clinical routine has been demonstrated difficult (11,16,17,19), there is a need to modify attitudes, cultivate new aptitudes, build specific DAs tools, and guarantee an ideal environment where communication and sharing viewpoints are appreciated (17,20,45,50,51).
The increasing number of publications about SDM and DAs in recent years shows that these topics are gaining relevance. Our systematic review revealed that LABC specific DAs were heterogeneous and scarce. We found very diverse instruments, from booklets to online tools, more or less interactive, standard or personalised; adapted or not to the reader's literacy,… They could be focused on the whole assistance process or a particular step or include a set of tools to be used sequentially as needed. Among this significant heterogeneity, we could underline the well-known Option Grids (20), which are brief summaries of alternatives with the most relevant questions that patients frequently ask, derived from common concerns with clear, concise and accessible language. These tools provide comparisons between alternatives and guide patients to being more willing to ask questions and pursue additional information, helping them become better informed to shared decisions with clinicians (20). Although this heterogenicity makes it challenging to assess and compare different DAs, reducing the possibility of showing conclusions that could be deduced, it indicates an urgent necessity for adequate guidance and consensus in their building. Our systematic review suggested that DAs for LABC care management facilitate understanding of the disease and decrease decision conflict and patient anxiety. In addition, it has been shown that the use of DAs allowed patients to undergo surgery when indicated, which could decrease patients' tendency to undergo treatment (29). Most of the DAs were focused on treatment. No DAs related to screening or follow-up were retrieved. This is unexpected because these are relevant features in the BC management process. It has already been shown that screening must be adapted to the characteristics (age, genetic factors, race, etc.), wishes and preferences of women (52). Concerning follow-up, BC should be treated as a chronic medical condition even in disease-free patients, so it should be necessary to develop DAs to increase knowledge and guide the patients through the process.
On the other hand, there are no established DAs to help with formal decisions in any aspect of the BC management process, and their development process is heterogeneous. There is still a long route before the achievement of consensus. Our work has recognised a gap and offers an essential contribution to directing further research and debate. Efforts must be made to develop integral and practical DAs to help patients to compare the different options available. Forthcoming investigations should be focused on studying easy Option Grids that could be more useful for the patients than leaving them alone with more complex interactive digital tools.

Conclusions
Implementing DAs is essential for high-quality patient-centred cancer care as it improves knowledge and reduces decision conflict in patients. This systematic review found that LABC DAs were heterogeneous and poor. There is a need for consensus in methodology and mandatory rules to facilitate specific DAs development and use to help patient decisions and enable SDM.

Declaration of interest
The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the article.

Data availability statement
The data and materials supporting the results are available from the corresponding author on reasonable request.

Funding
The author(s) reported there is no funding associated with the work featured in this article.