Shared decision-making for prostate cancer screening and treatment : a systematic review of randomised controlled trials

INTRODUCTION: Men facing prostate cancer screening and treatment need to make critical and highly preferencesensitive decisions that involve a variety of potential benefits and risks. Shared decision-making (SDM) is considered fundamental for “preference-sensitive” medical decisions and it is guideline-recommended. There is no single definition of SDM however. We systematically reviewed the extent of SDM implementation in interventions to facilitate SDM for prostate cancer screening and treatment.


Introduction
Prostate cancer is one of the most serious public health concerns relating to men's health worldwide.The World Health Organization (WHO) has declared prostate cancer to be the second most commonly diagnosed type of cancer in men, and the fifth leading cause of death due to cancer in men worldwide [1].It accounts for 6.6% of the total deaths of men, and the burden is expected to increase to 1.7 million cases and 499 000 new deaths by 2030 globally [2].Prostate cancer incidence varies widely in the world with higher rates (mostly) in high-income countries [1], mainly due to the widespread use of screening tests, which have improved early detection, but whose benefits and harms are controversial [3,4].There is no consensus on the general screening routine, including the age at which screening should be performed [5][6][7][8][9], and testing has led to false-positive results and over diagnosis [10].Furthermore, patients often face more than one alternative treatment, which represent a variety of benefits and risks without convincing evidence indicating a best choice [11].The survival benefit comes at the price of considerable morbidity, highly impaired quality of life, psychological distress and increased healthcare costs due to treatment [10,12].With these precedents, the individual patient's situation becomes preference sensitive, requiring careful consideration and deliberation of many factors (e.g., diagnosis, prognosis, fears, values, beliefs, ethics, hopes and previous experience) that make decisions complex and highly preference sensitive.Shared decision-making (SDM) is frequently advocated in clinical practice as the fundamental component of all patient-provider interactions in regards to medical decisions [13,14] since it is based on the principles of patient-cen-tred care [15,16].It is particularly recommended for "preference-sensitive medical decisions" [17] and considered essential for screening and treatment of prostate cancer [18,19].With this approach, the decision depends to a great extent on the patients' informed preferences and on their value of risks, benefits and harms of options [17].These attributes are often integrated and tailored to the patient's circumstance by means of decision aids or other methods [20][21][22][23] that facilitate SDM [16].However, there is no single definition of SDM and no clear consensus about how to conduct SDM in routine medical practice.Ongoing debate also indicates that the goal of SDM is not yet clarified.Some view SDM as a partnership between patient and/or patient care-related parties (e.g., legal guardian, relatives) and healthcare providers to equally share decisions about healthcare choices [24][25][26][27].For others, SDM is a process to engage in decision-making [14,28], or an approach to incorporate preference-sensitive elements that facilitate decision-making [17].SDM appeals greatly to policy makers and healthcare providers because of its potential to reduce the overuse of options with unclear benefits [29] while enhancing the use of beneficial options [30] and reducing variations in practice [31].We performed a systematic review to assess the extent of SDM implementation in studies of interventions aiming to facilitate SDM for men facing prostate cancer screening and/or treatment decisions.

Methods
We developed a protocol before starting the review following the principles for systematic reviews [32,33], and we report the methods in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines (see table S1 in appendix 1 for the PRISMA checklist) [34].

Inclusion and exclusion criteria
We focused on the extent to which the concept of SDM is implemented in clinical practice.We assessed the reported SDM interventions based on the SDM model (see section "Assessment of SDM implementation").We broadly defined SDM interventions as the approaches, methods or tools designed to facilitate, foster, or improve patienthealthcare provider involvement in medical decision-making, based on Charles et al. [35].We included peer-reviewed and grey literature of studies reported in English addressing (the effectiveness of) SDM interventions for men facing decisions about prostate cancer screening and/ or treatment.Eligible studies were randomised controlled trials (RCTs), quasi-RCTs (method of allocation not strictly random), and cluster RCTs (1) comparing SDM interventions to one or more alternative interventions, and/or usual care, (2) directed at patients and/or their care-related parties and/or healthcare providers, and (3) conducted in primary or specialised healthcare including general practices, community clinics, ambulatory care, hospitals and private care services.Studies were included regardless of the length of follow-up, publication year and country of origin.We excluded studies conducted in non-clinical settings and community studies in which discussions were not intended or could not occur.

Search strategy and data sources
We designed and conducted a comprehensive search strategy in Medline Ovid, Embase (Elsevier), CINHAL (EB-SCOHost), The Cochrane Library (Wiley), PsychINFO (EBSCOHost) and Scopus from their inception to March 2015.The search strategy was revised by an information specialist and, included terminology compatible with SDM (e.g., "patient participation" and "patient involvement"), "shared decision making" and "prostate cancer" (see table S2 in appendix 1).It was not restricted by publication date, language, country or outcomes, and included a study design filter for the identification of RCTs in humans [36].We also searched for grey literature using individual clinical trial registers (clinicaltrials.govand ISRCTN), the WHO search portal (http://apps.who.int/trialsearch), and the Ottawa Hospital Research Institute website (http://www.ohri.ca).The records were accessed between February and August 2016, and the trials registration number was additionally searched for by use of Medline and PubMed.We also used Google Scholar and the system for Information on Grey Literature in Europe (http://opengrey.eu/).We identified additional studies by screening the reference lists of included studies, relevant systematic reviews and clinical guidelines, and by contacting (June 2015 to January 2017) the authors of potentially eligible abstracts for which the full text could not be located.

Selection of studies
Two reviewers independently screened the titles and abstracts of all citations, and examined the full text of potentially eligible publications meeting the eligibility criteria.Studies reported in more than one publication were identified and treated as one unit.We resolved differences through consensus or by involving an arbitrator.

Data collection and synthesis
One reviewer extracted data using standardised data collection forms designed and developed a priori.A second reviewer independently verified data extractions, resolving differences by consensus or by involving an arbitrator.For each study, we extracted information on the bibliographic details of studies (design, country, time of study conduct, funding sources), characteristics of study populations and interventions, including the interventions' attributes, and the elements and key features of SDM implementation.Data from a single study reported across various publications were extracted as one unit.We obtained full-text data from the authors of potentially eligible abstracts without available full text.In this review, we performed a narrative synthesis of the results, including a description of the reported SDM interventions and their implementation based on the SDM model.In a future report, we will include an analysis of the effectiveness of SDM interventions.

Assessment of the extent of SDM implementation
We evaluated the extent of SDM implementation in accordance with the essential characteristics of SDM proposed by Charles et al. [35] (see table S3 in appendix 1).Of the analytic stages of SDM, we considered deliberation to be central and mandatory, and that it must be bi-directional (i.e., active participation of both patient and healthcare provider) for SDM to occur.Provision of information only, such as use of decision aids, cannot replace this active and bi-directional participation, but such strategies in a "standalone" format can facilitate SDM or become a component of a multi-faceted intervention.To differentiate the variants (e.g., two-way from one-way) in decision-making, we assessed the intervention's description and content, its delivery procedure and the mode of decisions to identify the elements aiming to facilitate decision-making.We evaluated whether: 1.The intervention aimed to facilitate or foster shared decisions, for example by including elements of patient activation, encouragement to talk or discuss, etc. 2. There was evidence of bi-directional interaction between patients and healthcare providers, such as planned (telephone or face-to-face) consultations.

Implementation of decision-making was based on
three key features of SDM [35], i.e., patient and healthcare provider: a. share/exchange information, b. deliberate, and c. make/implement a decision in consensus.
Ideally, this collection of behaviours occurs altogether within a clinical encounter [35].We anticipated, however, that SDM definitions and goals would differ among studies resulting in heterogeneous decision-making behaviours in which SDM might not be achieved.We classified the interventions as SDM (all criteria met), partial SDM (at least deliberation met), unclear (unclear deliberation), and no SDM (unidirectional interaction) by coding 3a, 3b and 3c as one if the criteria was met, zero if the criteria was not met, or unclear (?) if criteria details were not reported or could not be verified.Table S4 (appendix 1) illustrates this system.
We considered the following criteria as components of SDM, since these were intended to encourage discussions between patient and healthcare provider or implied a bidirectional interaction between them: patient activation strategies such as provision of information, patient prompts, clinical encounters that occurred at or shortly before a healthcare appointment, coaching, interviews, or before filling out questionnaires.

Identification of eligible studies
Our searches identified 15 398 records.After perusal of all titles and abstracts, we excluded 15 128 records.We examined in detail the full text of 270 potentially relevant articles.After excluding 220 articles, 36 RCTs reported in 50 publications met the inclusion criteria .Figure 1 shows the flow of study identification and  [82]; and two RCTs (5.6%) did not use age as an eligibility criterion for participants [68,74].Three RCTs were not tied to a consultation [38,48,57], but the type of participating healthcare providers was reported in 24 (66.7%)RCTs: 14 RCTs (38.8%) employed faculty, general or internal medicine physicians, and nurse practitioners; and 10 RCTs (27.8%) employed physician specialists (urology, oncology, and/or radiation oncology).Eleven (30.6%)RCTs reported the number of participating healthcare providers, which ranged from 2 [85] to 127 [54].Seven RCTs (21.2%) reported the level of healthcare providers' training or experience, which ranged from postgraduate practice to 40 years of experience, or board certified physicians.Thirty-four RCTs reported the funding sources; these were non-profit governmental and private institutions.

Attributes of decision-making interventions
The interventions varied widely in their delivery mode, form, and content (table 3).SDM was considered within the context of primary care in 55.5% (n = 20) of the RCTs, multidisciplinary healthcare in 19.4% (n = 7), hospital care in 14.0% (n = 5), specialised care in 8.3% (n = 3), and from a population perspective in 2.8% (n = 1).The interventions were delivered on-site (n = 14), home (n = 9), onsite or home (n = 9), home or on-site combined with other settings (n = 3), and face-to-face or by telephone (n = 1).[51]; in two RCTs interventions were designed for low health-literacy populations [46,54]; one RCT considered the target population with a literacy expert [58]; and one RCT used tailored literacy with a decision navigator [72].

Elements and key features of SDM interventions
Twenty-five RCTs (70%) intended to assess SDM to some degree (table 4).This intention was not clearly stated in the other 11 RCTs (30%), although the interventions included elements to facilitate or foster SDM in all but one study."Informed decision-making" was the most frequently (n = 21) used term, whereas only 9 (25%) RCTs used the term SDM.The studies also referred to other terms and measurements relevant to SDM including "weighing up benefits and harms", "risks", "pros and cons of options", "patients' values", "preferences", "promotion of engagement", "discussions of choices", "activation" or "participation in decision-making appointments", "decision role" (e.g., active, passive), "patient autonomy", "patient centredness", "knowledge and beliefs", and "decisional con  sion-making involved at least two parties, 45.2% (screening, n = 8; treatment, n = 6) fulfilled the three key SDM features: nine considered SDM within the context of primary care and five within the context of hospital and/or specialised care.Another 45.2% (screening, n = 10; treatment, n = 3; screening and treatment, n = 1) met the criteria for partial SDM (verified deliberation); 3.2% (treatment, n = 1) had all key SDM features difficult to verify (unclear deliberation), and 6.4% (screening, n = 1; treatment, n = 1) had the characteristics of no SDM.The other five (13.9%) of the 36 included RCTs, showed unclear de-liberation (screening, n = 1; treatment, n = 1) or no SDM (screening, n = 2; treatment, n = 1).

Discussion
In this systematic review, we identified 36 RCTs of interventions aiming to facilitate SDM for screening and treatment of prostate cancer in a variety of settings and populations.The majority of RCTs were from North America, mainly the USA (n = 22).Most of the participating men were 40 to 86 years old and more than half (55.6%) were recruited from primary care.There was a wide variation in  the minimum age (range: 40-55) at which men were targeted to be screened for prostate cancer with starting cutoff ages at 40, 45, 50, 55 years, and 18 years in one study.Primary care physicians or nurse practitioners participated in at least a third of the studies, whereas specialised physicians participated in less than a third of the studies.Most studies addressed decision-making for prostate cancer screening, with PSA being the most (78.3%)frequently used method of diagnosis.The interventions differed widely in delivery mode, format and content.Our approach for assessing the implementation of SDM interventions was based on the criteria defined by Charles et al. [24,35].The model distinguishes the roles and responsibilities of the relationship between patient and health-care provider for SDM compared with other models of decision-making.The essential characteristic of SDM is the bi-directional interaction between patient and healthcare provider which places SDM in the middle between a paternalistic and an informed-decision approach.Patients (and/ or related parties) and healthcare providers need to actively adopt a set of behaviours in each of the analytic stages, namely information exchange, deliberation and decision implementation [35].Our approach also supports deliberation as the key feature to accomplish SDM in routine practice, in keeping with Elwyn et al. [87].We found that different strategies are used to encourage participation in decision-making, and interventions might be considered to facilitate SDM, although they might not

Strengths and limitations
To our knowledge, this is the first systematic review about SDM implementation for both screening and treatment for prostate cancer.As such, this review focused on assessing and describing the reported SDM interventions and their implementation in clinical practice based on the SDM model.Given the lack of a single SDM definition, we considered the diversity in the type of interventions that would be compatible with SDM.Various reviews have focused on decision aids.We used a broad definition of SDM interventions and did not limit our search strategy exclusively to the term "shared decision-making" or "decision aids".We used a range of search terms relevant to decision-making, including SDM and decision aids.We applied broad inclusion criteria at the screening stage and full-text evaluation, and included studies regardless of whether a specific decision was promoted.Our review also covered international literature with no restriction to countries or type of healthcare provider.We included literature published in English only, and academic databases were searched up to March 2015.However, we made considerable efforts to identify all relevant studies by comprehensively searching both peer-reviewed and grey (accessed: February-August 2016) literature in twelve sources.We also contacted authors (2015-2017) of abstracts for which full texts were not available, increasing the chance of identifying more literature that is contemporary.Our work thus benefited from the response of authors, which led to the identification of more studies and thus more complete data were considered for eligibility.Moreover, our method for evaluating the implementation of SDM confirmed that research gaps in the conceptualisation of SDM continue despite previous recommendations [14].We used the SDM model by Charles et al. [35] because it represents only one SDM concept, and it is the most prominent [14] approach to viewing SDM compared with other models of decision-making.Our review thus presents the elements and key features of SDM  n.r.= not reported.General medicine = general, internal, family and/or community practice clinics, preventive medicine, Veterans' affair or primary practice clinics.Class: 1 = SDM, 2 = partial SDM, 3 = unclear deliberation, 4 = no SDM: no deliberation.Each SDM key feature [a-b-c] was coded as 1 = criteria met, 0 = criteria not met, or unclear (?) = judgement could not be made owing to unclear or lack of reporting (see table S4 in appendix 1).
interventions and provides an overview of the extent of SDM implementation for prostate cancer.
Our review was limited by the quality of reporting of intervention details, which made the verification of SDM criteria difficult at times.Thus we cannot exclude the possibility that we underestimated SDM implementation.Many studies were published within the last decade, but the use of frameworks was lacking in nearly a third of them.

Conclusions
There is a significant variation in the components of SDM interventions for prostate cancer screening and treatment.Only 39% of the studies contained the SDM intervention components suggested in the SDM model, and interventions were implemented mostly within the context of primary care.These results merit further evaluation on patient outcomes.There might be strong ethical, medical and interpersonal reasons to recommend SDM.However, to date there seems to be uncertainty about the SDM concept, intervention content, and how to implement SDM in practice.A standardised SDM definition and guidance for SDM implementation in practice that is feasible for several clinical settings are needed.

Figure 1 :
Figure 1: Identification and selection of studies.

Table 1 :
Summary of the characteristics of 36 randomised controlled trials of decision-making interventions for prostate cancer.

Table 2 :
Characteristics of 36 randomised controlled trials of decision-making interventions for prostate cancer.
Swiss Medical Weekly • PDF of the online version • www.smw.chPublished under the copyright license "Attribution -Non-Commercial -No Derivatives 4.0".No commercial reuse without permission.See http://emh.ch/en/services/permissions.html.

Table 3 :
Characteristics of decision-making interventions for prostate cancer screening and treatment.
[14]s Medical Weekly • PDF of the online version • www.smw.chPublishedunder the copyright license "Attribution -Non-Commercial -No Derivatives 4.0".No commercial reuse without permission.See http://emh.ch/en/services/permissions.html.anceforSDM implementation in routine practice.Makoul et al.[14]identified a range of 31 different SDM definitions and, as noted in our review, their recommendations for a single and more integrative concept of SDM are yet to be followed.Future research should consider that this variability might make comparison across studies difficult, and that consistent reporting of interventions and their compo-

Table 4 :
Elements and key features of decision-making interventions for prostate cancer screening and treatment.
Swiss Medical Weekly • PDF of the online version • www.smw.chPublished under the copyright license "Attribution -Non-Commercial -No Derivatives 4.0".No commercial reuse without permission.See http://emh.ch/en/services/permissions.html.