Shared decision making in surgery: A scoping review of the literature

Abstract Background Shared decision making (SDM) has been increasingly implemented to improve health‐care outcomes. Despite the mixed efficacy of SDM to provide better patient‐guided care, its use in surgery has not been studied. The aim of this study was to systematically review SDM application in surgery. Design The search strategy, developed with a medical librarian, included nine databases from inception until June 2019. After a 2‐person title and abstract screen, full‐text publications were analysed. Data collected included author, year, surgical discipline, location, study duration, type of decision aid, survey methodology and variable outcomes. Quantitative and qualitative cross‐sectional studies, as well as RCTs, were included. Results A total of 6060 studies were retrieved. A total of 148 were included in the final review. The majority of the studies were in plastic surgery, followed by general surgery and orthopaedics. The use of SDM decreased surgical intervention rate (12 of 22), decisional conflict (25 of 29), and decisional regret (5 of 5), and increased decisional satisfaction (17 of 21), knowledge (33 of 35), SDM preference (13 of 16), and physician trust (4 of 6). Time increase per patient encounter was inconclusive. Cross‐sectional studies showed that patients prefer shared treatment and surgical treatment varied less. The results of SDM per type of decision aid vary in terms of their outcome. Conclusion SDM in surgery decreases decisional conflict, anxiety and surgical intervention rates, while increasing knowledge retained decisional satisfaction, quality and physician trust. Surgical patients also appear to prefer SDM paradigms. SDM appears beneficial in surgery and therefore worth promoting and expanding in use.


| ME THODS
The protocol for the study was designed in concordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) checklist and registered with PROSPERO (CRD42018097286), the international registry for systematic reviews. 13 Necessary to the search was an operationalized definition of SDM, as occasionally the term has been used incorrectly. 15 The following working definition was determined most appropriate: 'an approach where clinicians and patients share the best available evidence when faced with the task of making decisions, and where patients are supported to consider options and to achieve informed preferences'. 16 Such a definition was used by 1 in the most comprehensive SDM review of medical specialties. 16 Moreover, it is fitting for surgery, where best evidence may require other paradigms besides SDM to be used, such as in the case of emergency surgery. 17 Given the definition, competing SDM frameworks were not included in the search. This resulted in some papers being excluded based on other forms of patient-physician relationship, such as informed consent only. 18

| Inclusion and exclusion criteria
Publications were included if they discussed SDM by surgical discipline. Both emergency and elective procedures were included. Studies were excluded if they did not contain explicit methods for quantifying SDM; if they were commentaries, opinion pieces, reviews, committee reports, or conference abstracts; and if surgery was not made the explicit outcome in the abstract. Studies were also excluded if they discussed other means of shared patient information, such as genetic screening. This resulted in including both qualitative and quantitative cross-sectional studies, as well as randomized controlled trials.
Two investigators (KN and SM) independently reviewed all eligible titles and abstracts. In instances of disagreement, a third reviewer (DP) adjudicated. A full-text screen was then performed to determine acceptability into the final review ( Figure 1). This was done by KN to ensure inclusion criteria were met, if the studies used SDM in a surgical discipline only, and to understand how the studies differed in their use of SDM.

| Data extraction and quality assessment
Data were extracted from two types of publications: controlled comparative studies and exploratory cross-sectional studies. For each, the author(s), year, surgical discipline, total number of patients, number of sites, location, median patient age, study duration, whether a decisional aid was used, intervention type, survey methodology and variable outcomes were recorded (Table 1). For the controlled comparative studies, the total number of participants in controls compared to the intervention group was recorded. For the cross-sectional studies, additional qualitative comments and themes were noted. This included two authors (KN and EG) to categorize the papers individually. The two identified, coded and grouped relevant text using Excel (4) . The common words were grouped by frequency and similarity post hoc, in a method similar to the Joanna Briggs Institute guidance. 19 The quality of each RCT was analysed manually using the Cochrane Bias Risk Assessment tool (Table 2). RCTs were additionally grouped based on the type of decisional aid used, the study design type (preoperative or postoperative), and the surgical speciality.
The observational studies were not coded using an assessment tool due to their heterogeneity.  Table 1 illustrates the general characteristics of the final studies. The majority of the publications (67%) were single-institution studies, with most (67%) being cross-sectional. Forty-eight per cent of the studies had a small sample size (less than 100 patients), 37% focused on patients aged 50-65 years, and 69% originated in North America. The most represented specialty was plastic surgery, particularly focusing on the choice between mastectomy and nonsurgical therapy for breast cancer ( Figure 2). This was followed by general surgery (for elective surgeries) and orthopaedics (primarily osteoarthritis), paediatric surgery with 3 studies, and neurosurgery with only one qualitative cross-sectional study.

| RE SULTS
The publications selected included 99 qualitative and quantitative cross-sectional studies and 49 RCTs. Decision aids were used in 91 RCTs and cross-sectional studies, with the others using SDM models to elicit patient preferences. Seven types of decisional aids were employed: paper/booklet format (34.5%), web-based (20.2%), coaching (16.7%) and mixed media that used more than one type of decisional aid (13.1%), including 80% video with paper, 10.7% video alone, 5% with a multicomponent of coaching and a paper, 3.6% mobile app, and 1.2% object-based.

F I G U R E 1 Identification of eligible studies in scoping review
Two themes emerged from the review: measurable outcomes from the surgical DAs, and qualitative descriptions of SDM from cross-sectional studies. The primary outcomes measured using surgical DAs were intervention rates following SDM, decisional conflict in comparative groups, and the time required using an SDM paradigm. 20 Secondary outcomes including decisional satisfaction, decisional regret, knowledge provided, decisional anxiety and quality, and increased patient-physician trust were also aggregated according to their definitions in the literature. 21 These selected outcomes are the most common outcomes underpinning proper SDM paradigms. 22 Figure 3 summarizes the specific characteristics of the DAs.
The majority of the studies were of good quality, although detection and performance biases were a concern.

| Surgical rates & increased knowledge
Twenty-two of the 91 studies using DAs discussed the rates of surgery versus alternative therapies such as active surveillance. Some studies (n = 7) noticed an increase in surgical rates, particularly when tied to an increase in knowledge and when dealing with minority populations. [23][24][25] The majority (n = 12) however noted a decrease in chosen surgical rates compared to more conservative treatment methods. This decrease in surgery was mirrored in the cross-sectional studies. Where there was an increase in knowledge, patients in some of the studies (n = 2) determined that their knowledge of alternative treatments was also bolstered. Such an increase in knowledge, often resulting in a lower threshold of surgical pain expectation, was also coupled to decreased intervention rates. Other studies have noted an increase in physician-patient trust (n = 3), even if there was a decrease in the number of surgical interventions.

| Decisional conflict
In 86% of the publications, decisional conflict was noted to decrease (n = 25) in intervention arms when compared to controls, with only two studies recording no difference between the control and study arms. This decrease was paralleled by a similar increase noted in decisional satisfaction (n = 9) when decisional conflict decreased. Four studies recorded decision quality, defined as the ability to make decisions aligned with the best information and patient values. 26 Nearly all papers (n = 3) saw an increase in the quality of decision made.
In only one study was there an increase in decisional conflict in the intervention arm using SDM. Similarly, where recorded, decisional anxiety was seen to decrease (n = 6). Decisional regret was also observed to decrease in the 5 studies using DAs.

| Time spent and SDM preferred
Six papers recorded how much time was spent with SDM models ( Figure 3). It was noted that time was seen to increase in three, decrease in two, and not change in one. Time was seen to increase on average by 10-30 minutes.
In each of these papers, it was also noted that there was an overwhelming support for SDM, recorded as SDM preferred over the usual treatment regimens, despite increased time spent in decision making. Of the studies that compared rates of SDM preference (5 of the 9), the increase in preference for SDM was high (range of 13% to 35%), whereas only a few (3 of 9) noted no difference in preference, with one noting a decrease in SDM preference.   Mobile apps and object-based aids did not include enough studies that recorded the associated variables.

| Narrative characteristics
Of the studies that did not specifically use a DA (n = 55) but still explored SDM in surgical disciplines, whether through qualitative surveys of departments or patient interviewing, trends emerged in the patients' perception and expectation for particular care. While a large majority of the studies were focused on the comments from small samples (n = 28), a comparison between SDM and non-SDM explorative studies was possible (Table 3). In studies that did not define their SDM protocol but rather explored the idea of using SDM in their select populations, there was a large degree of treatment

| D ISCUSS I ON
Health care has moved towards patient-centred approaches. 27 With its widely studied use in screening, chronic conditions and cancer diagnosis, SDM is generally considered among the best current means to provide optimal treatment outcomes. [1][2][3][4] Many of the same driving effects of SDM were found in this review of surgical disciplines.
We have observed that, generally, surgical rates and decisional conflict decreased, whereas knowledge obtained increased. Time spent with the patient, on the other hand, was inconclusive. This was true across all types of DAs used in SDM, though certain types (such as in-person coaching) had a more consistent effect on specific outcomes such as decisional conflict. Narratively, a similar theme has been observed: patients prefer SDM, believing that their health care becomes more accessible, and treatment goals were aligned compared to those without SDM. These gains are reinforced by the uniform preference of patients undergoing surgery for SDM, a care that aligns with their values and preferences, and allows them to make the most informed decision in their care, especially when alternative treatment methods exist.
These results are both encouraging and consistent with prior literature. 28,29 Other reviews have also recorded a decrease in surgical rates, particularly as they relate to breast cancer interventions. 30 This result is however only beneficial if the risks of not having surgery are outweighed by a patient's understanding and acceptance of both their current and future conditions. 31 While our study did not deal with this aspect, they note that more conservative therapy appears to be preferred with SDM.
Of those studies that showed an increase in choosing surgical interventions, the effect was generally intended. Such for instance is a study in which the DA was tailored specifically to increase arthroplasty rate in racial minorities' populations. 23 This supports a well-documented aspect of SDM research-racial minority and ethnic minority populations may benefit more than others from SDM paradigms. 24,25 Race/ethnicity of study participants was not reported, and therefore, any variation in the effect of the intervention across this demographic could not be ascertained. Other social determinants of health, like socio-economic or education status, were also not recorded. As a result, the exact effect could not be quantified.
This is true of other patient characteristics too, such as the fact that many of the populations studied were older, with only one SDM protocol for paediatric populations. In those older populations, the knowledge of surgical procedures is largely considered lower, particularly for more advanced procedures. 32 DAs may be a beneficial way to bridge this knowledge deficit, especially if SDM provides a reinforcing effect in minority populations. DAs would therefore provide a standardized way across the surgical disciplines to formalize care. Rather than vulnerable patients finding themselves voiceless in front of influential surgeons' advice, SDM paradigms tend to equalize the patient-provider power balance within the therapeutic relationship, thereby, as observed, increasing physician trust.
In light of this effect of equalizing practitioners and patient, especially in minority populations, SDM has been lauded by the American Medical Association (AMA) as the future of medical practice. 17 36 Our study also does not investigate comparative care regimes that define controls differently for the same procedure. This limitation is further exacerbated by the fact that the outcomes were often measured by author-defined means, rather than using validated scales, such as the decisional conflict or the decisional regret scale. 36,37 This may be intentional, given the lack of rigor and standard of using SDM outcomes s in surgical disciplines. 37 Finally, this study notes the incidence of SDM in various surgeries but cannot describe their varied use in any one surgery, as well as the outcomes of surgeries (such as less complications or infections). This is partly due to the limited spectrum of subspecialties. Most SDM studies within surgery were conducted in plastics, followed by general surgery and orthopaedics. Therefore, any extrapolations on other surgical specialties need to be guarded. This is also due to their varied use and definitions of SDM, even within the same discipline, as well as small sample sizes of each.
Despite these limitations, the implications for clinical practice are significant. If SDM can provide timely, appropriate care management for patients that coincides with their own values, then the possibility of undesired outcomes may be decreased. In particular, this research suggests that decisional regret will go down, knowledge regarding treatment modalities, and that a better therapeutic relationship would develop. Moreover, from the narrative review, patients appear to be preferring SDM, even in disciplines like surgery.

| CON CLUS ION
SDM is advocated for providing the utmost efficacious, reliable and patient-focused care. This systematic review shows that the same principles guiding SDM in general also ground SDM in surgical settings. These findings can encourage the further application of SDM in surgery, especially for particular contexts such as elective procedures or in instances of equipoise where SDM is an extension of true informed consent. 35,38 Further studies would need to explore the types of DAs in their respective surgical fields, as well as the reasons for the wide variability in SDM penetration across surgical specialties. The potential unique application of surgical SDM for minority populations such as children, the elderly, and ethnic and racial minorities also warrants further study.

DATA AVA I L A B I L I T Y S TAT E M E N T
The authors confirm that the data supporting the findings of this study are available within the article [and/or] its supplementary materials.