Shared decision making for men facing prostate cancer treatment: a systematic review of randomized controlled trials

Introduction

Prostate cancer (PC) is the second leading cancer in men and the fifth leading cause of death due to cancer in men worldwide.^1,2 Patients with PC often face more than one alternative to treatment eg, active treatment, active surveillance, or watchful waiting. These choices involve trade-offs between benefits and harms due to the limited evidence regarding the optimal treatment strategy for PC.^3–5 The survival benefit of treatment options including surveillance is associated with considerable morbidity due to potential adverse outcomes of treatment (eg, urinary and erectile dysfunction, loss of fertility, and chemotherapy and/or hormone therapy side effects), psychological distress, and impaired quality of life.^3–6 These factors make treatment decisions complex and highly preference-sensitive. Patients thus need to weigh carefully not only the diagnoses and prognoses but also their own fears, values, beliefs, ethics, hopes, and previous experience.

Shared decision making (SDM) is viewed as an approach to involve patients and their clinicians in a process of collaboration and deliberation to reach medical decisions, particularly for preference-sensitive conditions.^7–9 SDM helps inform patients about the options for, and the effectiveness of, treatment, taking into account the patient’s needs, knowledge and their value of risks, benefits, and harms. Health authorities and policy makers strongly encourage SDM for decisions about PC treatment.^10–13 There is variation in the level of SDM implementation however, mainly due to the lack of consensus in SDM definition and goals.¹⁴ We performed a systematic review to evaluate the evidence for the effectiveness of SDM compared to current clinical practice for the treatment of PC.

Methods

We followed a protocol based on the principles for systematic reviews^15,16 and report the methods according to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines (see Table S1).¹⁷ We identified and selected studies following the methods published in detail elsewhere.¹⁴

Eligibility criteria

We included peer-reviewed and grey literature from RCTs published in English comparing the effects of SDM to usual care for decisions about PC treatment, which were conducted in primary and specialized care (general practice, community clinics, ambulatory care, hospital and private). We further limited the inclusion of studies to RCTs that, regardless of the intervention target (patients, HCP, surrogates, or family members): 1) met the criteria for SDM^9,18 supporting the principle of bi-directional deliberation,⁸ as previously illustrated,¹⁴ and 2) reported data in quantitative format for the outcomes of knowledge, patient satisfaction, perception of being informed, risk perception, decisional conflict, decisional regret, outcomes of emotional distress caused by the disease itself and/or treatment (eg, mood disturbance, anxiety, and depression), treatment behavior (eg, adherence to treatment), quality of life (QoL), symptoms, and mortality.

Study identification and search strategy

We searched for studies in: 1) academic databases: MEDLINE (Ovid), EMBASE (Elsevier), CINHAL (EBSCOHost), The Cochrane Library (Wiley), PsychINFO (EBSCOHost) and Scopus up to March 2015 (see Table S2); and 2) grey literature records (accessed: Feb–Aug, 2016) from clinicaltrials.gov and ISRCTN, the WHO search portal (http://apps.who.int/trialsearch), the Ottawa Hospital Research Institute website (http://www.ohri.ca), Google Scholar, and the system for Information on Grey Literature in Europe (http://opengrey.eu/). We additionally searched for the registration numbers of the trials using Medline and PubMed. We screened the reference lists of included studies, relevant reviews, and clinical guidelines, and contacted (Jun 2015–Jan 2017) the authors of abstracts for which the full-texts could not be located.

Selection of studies and data extraction

Two independent reviewers screened all titles and abstracts, and assessed the full-text of eligible publications. One reviewer extracted all data and a second reviewer independently verified data extractions. Differences in study selection and data extractions were resolved by consensus or by involving an arbitrator. We grouped outcomes into affective-cognitive, behavioral, and health outcomes following the system by Shay and Lafata (2015).¹⁹

Risk of bias

Two reviewers independently assessed in duplicate the quality features of the included studies following established guidelines and resolving differences by discussion.^16,20 We rated the adequacy of core items including generation of random sequence, concealment of allocation at randomization, blinding (patients, health care providers, and outcome assessors), intention-to-treat (ITT) (if participants were analyzed based on their original group allocation), follow-up (same length of time), and if there were attrition rates of significant concern (at least 20%). We also rated whether studies reported on the definition of inclusion and exclusion criteria, “a-priori” sample size calculation, primary and secondary outcomes, and funding sources.

Data synthesis

There was mostly one study per outcome precluding the ability to perform meta-analyses. We thus analyzed the data based on individual trial estimates. Where data were sufficiently reported, we calculated the unadjusted risk ratios (RR) or the standard mean differences (SMD) and the 95% confidence intervals (CI) assuming a random-effects model in RevMan 5.3.5 software.²¹

Results

We examined in detail the full-text of 270 articles. Four of these fulfilled the SDM criteria and investigated the comparative effectiveness of SDM with usual care (Figure 1).

Figure 1 Process of identification and selection of studies.Abbreviations: GDL, guidelines; SR, systematic reviews; SDM, shared decision-making.

Study and population characteristics

The studies were from Canada (n=3) and the USA (n=1), and comprised 1,065 patients individually randomized to the intervention groups (Table 1). The patients were recruited from multidisciplinary or specialized care in hospital or general medicine. Three RCTs reported on cancer staging: 89.8% of the men had localized PC (clinical stage T1-T2),^22–24 6.2% were in advanced stage (T3-T4),²³ and 4.1% were of “unknown” clinical stage.^22,23 Radical prostatectomy (36.7%) and watchful waiting (28.5%) were the most frequently selected treatments among 743 men from 2 RCTs, 1 in general (hospital) and 1 in specialized cancer care, respectively. “No treatment” accounted for 25.4% of all options offered in one study.²³ Men had a mean age of 64.6 (SD 7.8), 83.9% were married and at least 53.4% had a minimum of high school education. The only RCT reporting on ethnicity included White (71.5%) and Black (28.5%) men. In three RCTs reporting on employment status, 55.3% of the men were in full- or part-time employment. In two RCTs, 88.2% of the participating health care providers were radiation oncologists and 11.8% were urologists.

Table 1 Characteristics of included studies

Table 1 (Continued).

Table 1 (Continued).

Interventions' characteristics

All interventions fulfilled the key features of SDM as illustrated in detail elsewhere.¹⁴ All RCTs used patient-directed interventions; one included the patients’ partners in the interventions sessions (Table 1).²⁵ The interventions were delivered before decision making, within the time of scheduled visits²³ or before consultations^22,24,25 in order to empower patient participation in decisions. Men were recommended to review the material before consultation or were specifically encouraged to participate in treatment decisions. Men were advised to bring their significant other(s) to the consultation in one RCT.²⁵ Three RCTs used multifaceted interventions on-site^22,25 or at home.²⁴ The interventions were self-administered or delivered by the research staff or by a nurse. The formats of the interventions included video, printed paper-based material, interviews, telephone calls, or audiotape recording. Only one RCT considered health care literacy for the development or pilot testing of the interventions.²⁴ One RCT evaluated two SDM interventions and usual care.²⁴ The content of the interventions included educational information, eg, about PC, treatment choices, advantages and disadvantages of treatment, side effects, and prognosis.

Risk of bias in the methods of included studies

All trials adequately randomized patients, but only two reported adequate concealment of allocation, leading to risk of selection bias in the other trials (Table 2). Only one RCT blinded patients and physicians, another blinded patients only, and no study performed or reported on the blinding of outcome assessors, resulting in high risk of performance and detection biases. Two RCTs with an attrition rate of less than 20% did not report on ITT analyses. Two RCTs had unclear reporting of attrition rates and ITT techniques. Men in all intervention arms were followed-up for the same length of time. All trials reported the participants’ inclusion criteria but only one trial reported the exclusion criteria too. Two RCTs reported on the sample size calculation and power. All RCTs measured the success of interventions by definition of primary outcome(s). All RCTs were funded by non-profit organizations.

Table 2 Methodological features of included studies

Effectiveness of interventions

There were twenty-four outcome measures reported in the four RCTs (Figure 2). Data were sparsely reported with one study per outcome, limiting the ability to conduct meta-analyses. Table 3 shows the effect estimates for each individual trial.

Figure 2 Outcomes reported in the included studies.Notes: Green = quantitative data; Yellow = qualitative data; Red = no outcome data.Abbreviation: DM, Decision-Making.

Table 3 Outcome effect estimates for SDM compared to usual care for decisions about prostate cancer treatment

Table 3 (Continued).

Table 3 (Continued).

Table 3 (Continued).

Table 3 (Continued).

Table 3 (Continued).

Affective-cognitive outcomes

Knowledge

One trial implemented two SDM interventions.²⁴ The different components between SDM groups consisted of nurse telephone calls to patients (treatment direct [TD]) or nurse telephone calls to patients and primary supporting persons (treatment supplement [TS]). Compared to usual care, TD significantly improved knowledge for PC and treatment at four weeks (SMD 0.33, 95% CI 0.03 to 0.64, p=0.03) and at three months (SMD 0.35, 95% CI 0.04 to 0.66, p=0.02). TS also showed a small but significant improvement in knowledge at four weeks (SMD 0.33, 95% CI 0.02 to 0.64, p=0.04), but not at three months. The combined effect of both SDM (TD and TS) interventions showed a significant improvement in knowledge at four weeks (SMD 0.35, 95% CI 0.08 to 0.62, p=0.01) and at three months (SMD 0.31, 95% CI 0.04 to 0.58, p=0.02).

Patient satisfaction

Individual trial effect estimates showed no significant difference between SDM and usual care in the number of patients who were satisfied with their treatment choice or who were satisfied with the levels of involvement in treatment decision making with the doctor.²² Scores of patient satisfaction with preparation for decision making were not significant between groups.

Perception of being informed and risk perception

Individual trial effect estimates showed a small but significant effect of SDM when compared to usual care on patients’ perception of receiving information (SMD 0.20, 95% CI 0.01 to 0.39, p=0.04).²³ The same trial showed a marginal but statistically significant effect of SDM on positive perceptions of receiving information about treatment alternatives (SMD 0.19, 95% CI −0.00 to 0.38, p=0.05) and treatment side effects (SMD 0.25, 95% CI 0.06 to 0.44, p=0.010). No study reported on risk perception.

Mood state/disturbance, anxiety, and depression

Outcomes of emotional distress including mood/disturbance, anxiety, and depression caused by the personal situation, the disease itself, or treatment were scarcely reported. Individual trial effect estimates showed no significant differences between SDM (TD or TS) and usual care with respect to mood state or disturbance at four weeks or at three months.²⁴ No study reported quantitative data on anxiety and depression.

Decisional conflict

Individual trial effect estimates showed no significant differences between SDM and usual care in the scores of decisional conflict.²²

Decisional regret

Individual trial effect estimates showed no significant differences between SDM (TD or TS) and usual care in the scores of decisional regret at three months.²⁴ The combined effect of both SDM (TD and TS) interventions compared to usual care was not significant at three months.

Behavioral outcomes

Adherence to treatment of choice

Radical prostatectomy and watchful waiting were the most frequent treatments received or chosen by men in two RCTs.^22,23 No trial reported on the adherence to a treatment initially chosen however.

Health outcomes

Quality of life, symptoms, and mortality

Individual trial effect estimates showed a significant effect of two SDM (TD or TS) interventions, compared to usual care, on men’s positive feelings about their QoL at four weeks (TD: SMD 0.50, 95% CI 0.19 to 0.81, p=0.002; TS: SMD 0.50, 95% CI 0.18 to 0.81, p=0.002).²⁴ This effect did not sustain at three months for either of the SDM (TD or TS) interventions however. Health outcomes of symptoms and mortality were not reported.

Discussion

In this systematic review, we identified four RCTs that both fulfilled the criteria for SDM and evaluated the comparative effectiveness of SDM with usual care for men faced with decisions about PC treatment.

Despite the low volume of literature, 37.5% of the study estimates significantly favored SDM when compared to usual care. We found moderate effects of SDM on knowledge, perception of being informed, and QoL. The effects on knowledge and QoL did not sustain at long term (three months) however. The remaining 62.5% of the study estimates showed no significant differences between SDM and usual care on patient satisfaction, decisional conflict, decisional regret, and mood.

This systematic review is based on published data. Surprisingly, six of the thirteen outcomes of interest were not reported. Of particular concern is the dearth of outcome data on the adherence to treatment and health outcomes including symptoms, mortality and QoL. The methodological quality of studies is low to moderate at best, mainly due to unclear allocation concealment, blinding and attrition. The interventions varied in characteristics and content, and were delivered before decision making mostly by trained nurses or research staff. The description of population characteristics was often incomplete too. The studies were carried out mostly in specialized and hospital care in high-income countries from North America, mainly Canada. The ethnicity of the populations was generally not described. Thus, the evidence is generalizable mostly to middle-aged men of at least fifty-five years of age, from Western countries, married, with low to moderate levels of education, who have an English-speaking background and face decision making for PC treatment.

SDM is highly recommended by major task force associations, policy makers, and clinical guidelines for medical decisions regarding PC treatment.^{10,11,26–30} Our systematic review, however, reveals a lack and variable reporting of outcome measures in the included studies, particularly on health outcomes. This makes it difficult to relate patient involvement in decision making to the actual effects of SDM. The inconsistent reporting of outcomes across studies is most likely due to the lack of consensus in the definition and in the approach used to evaluate and implement SDM.¹⁴ SDM is variably adopted in practice as suggested by the very few studies meeting the key SDM criteria. Perspectives of the health care team involved in SDM may play a role in this variability. In a multidisciplinary survey study, 63–71% of oncology nurses, urologists, and oncologists agreed that patients should be involved in decision making.³¹ However, 52–55% of the urologists and oncologists felt inadequately trained to apply SDM in clinical practice. On the other hand, 20% of the oncology nurses felt inadequately trained to apply SDM. Although an assessment of economic outcomes was outside the focus of our review, we noted that no study reported on cost data.

Clinical implications

The variation in the definitions and goals of SDM¹⁴ and the inconsistent reporting of outcomes across studies are the main barriers in generating an evidence base for the effectiveness of SDM in the care of PC patients. Before implementing SDM in clinical guidelines of PC care and applying these on a large scale in practice, more efforts in standardizing the definition, goals, and outcomes of SDM are needed. Our results may thus encourage researchers to engage in this area. The absence of evidence of SDM in PC care does not prove a lack of its beneficial effects. Thus, our results about the moderate effects of SDM on knowledge, perception of being informed and QoL may still motivate clinicians to activate their PC patients to participate in decision making.

Future research

Future research warrants further focus on the use of SDM interventions for decisions about PC treatment. Some guidance for SDM implementation based on a clear definition and objectives of SDM as previously reported¹⁴ could be the leading step in building focused and solid evidence on SDM. Agreement on a standard set of outcomes that are best to assess SDM and that are most meaningful to patients is critical to guide appropriate outcome collection and evaluation of SDM. Future research then needs to address the absence of evidence, particularly on health outcomes. The link between patient involvement in decision making and the effect of SDM-chosen treatment on patient-important and health-related outcomes needs special consideration. Since treatment can have a significant impact on the patients’ QoL and length of survival, studies of SDM for PC need to provide an appropriate description of the characteristics of the populations including comorbidities, status of cognitive function, literacy levels, sexual health, religious beliefs, side effects of treatment, and whether significant persons (eg, carer, partners) accompany the patients in the process of care. These factors influence not only the type of treatment of choice but also the patients’ emotions, decisional regret, and the degree of involvement in decision making.³² Patients with PC aged <40 years, for example, express significantly high positive and negative emotions, and partners of PC patients express more negative emotions than the patient himself. The physicians’ specialty may also unintentionally influence the physicians’ preference for treatment. Future studies should consider and address the perspectives of the health care team on SDM as a potential barrier for its implementation.³¹ In particular, interventions and components should be described in detail so that results are interpreted appropriately, and interventions can be replicated and evaluated. SDM should also be evaluated in relation to its costs and health gains so that SDM interventions can be reliably implemented. Therefore, future research needs to address the absence of evidence on cost data.

Strengths and limitations

To our knowledge, this is the first systematic review on SDM compared to usual care for men faced with decisions about PC treatment. It benefits from the inclusion of international literature without restrictions on countries or type of health care professionals. Including RCTs allows the estimation of causal effects with lower risk of bias. We only included studies meeting the key features of SDM^9,14,33 because of the continuing gaps in the conceptualization and implementation of SDM. We considered studies regardless of whether a specific decision was promoted. This was a rigorous and focused approach although we cannot exclude the possibility of underreported SDM characteristics in other studies. We only included literature published in English, but we made considerable efforts to identify all relevant studies. We searched several sources and contacted, between 2015 and 2017, the authors of relevant abstracts with no available full-text, thus increasing the likelihood of identifying more recent literature. The results from our systematic review are limited by the quantity and methodological quality of the available literature, in particular by a dearth of outcome data, and by lower quality reporting of outcome data. The studies focus primarily on SDM process-related outcomes and lack important data on health outcomes, affecting the ability to conduct meta-analyses.

Conclusion

There is little research currently available to appropriately evaluate the presumed benefits of SDM for decisions about PC treatment when compared to usual care. SDM may improve knowledge and perception of being informed and may have a positive but short-term effect on patient-perceived QoL. The (long-term) effects of SDM are unclear. Future rigorous research needs a consistent and relevant set of outcomes and interventions to assess the effects of SDM. In particular, it should address the absence of evidence and appropriately describe and test a reproducible form of SDM.