Do Decision Aids Benefit Patients with Chronic Musculoskeletal Pain? A Systematic Review

Abstract

Objective

To review the effect of patient decision aids for adults making treatment decisions regarding the management of chronic musculoskeletal pain.

Methods

We performed a systematic review of randomized controlled trials of adults using patient decision aids to make treatment decisions for chronic musculoskeletal pain in the outpatient setting.

Results

Of 477 records screened, 17 met the inclusion criteria. Chronic musculoskeletal pain conditions included osteoarthritis of the hip, knee, or trapeziometacarpal joint and back pain. Thirteen studies evaluated the use of a decision aid for deciding between surgical and nonsurgical management. The remaining four studies evaluated decision aids for nonsurgical treatment options. Outcomes included decision quality, pain, function, and surgery utilization. The effects of decision aids on decision-making outcomes were mixed. Comparing decision aids with usual care, all five studies that examined knowledge scores found improvement in patient knowledge. None of the four studies that evaluated satisfaction with the decision-making process found a difference with use of a decision aid. There was limited and inconsistent data on other decision-related outcomes. Of the eight studies that evaluated surgery utilization, seven found no difference in surgery rates with use of a decision aid. Five studies made comparisons between different types of decision aids, and there was no clearly superior format.

Conclusions

Decision aids may improve patients’ knowledge about treatment options for chronic musculoskeletal pain but largely did not impact other outcomes. Future efforts should focus on improving the effectiveness of decision aids and incorporating nonpharmacologic and nonsurgical management options.

Introduction

Chronic musculoskeletal (MSK) pain affects more than half of adults over the age of 18 in the United States. It is even more prevalent in older adults, affecting 61% of adults between the ages of 65 and 74 years, and 72% of adults >75 years [1]. Among US adults, low back pain is the most commonly occurring MSK disorder and is the leading cause of years lived with disability globally [2,3]. In 2011, the estimated annual cost of MSK-associated health care was $796.3 billion, 5.2% of the national gross domestic product [4]. Moreover, the prevalence and economic impact of MSK disease will continue to increase as the population ages [1].

Older adults with chronic MSK pain have varying preferences for their level of participation in health care decision-making, with many desiring an active role but some preferring to defer decision-making to their health care provider [5]. Regardless, older adults value being informed and discussing options with their provider [5]. Despite preferences for information, research suggests that patients are not well informed about treatment options for chronic MSK pain [6–8]. Older adults rely on information from their provider but also seek decision support from social contacts who have chronic pain [9–11]. When older adults were provided explicit information about the risks and benefits of various treatments for knee osteoarthritis (OA), their treatment preferences differed from common provider prescriptive practices [12]. This suggests that there is room for improvement in the decision-making process surrounding the treatment of chronic MSK pain, particularly in older adults.

The provision of patient-centered care is one of the six aims for improving health care quality in the United States outlined in the 2001 Institute of Medicine “Crossing the Quality Chasm” report [13]. Most treatment decisions for chronic MSK pain are preference-sensitive, as they include multiple reasonable treatment options [14]. Further, many currently available treatments have similar efficacies but vary significantly in terms of side effect profiles and cost [12,15]. For these preference-sensitive decisions, it is important to ensure that patients make an informed decision that is congruent with their values [16].

Patient decision aids are tools designed to help patients learn about their treatment options and participate in their health care decisions. They can take many forms, including paper booklets, videos, and computer- or technology-based applications. Decision aids explain the decision being made, present both positive and negative features of each option, and help patients clarify their personal values. Importantly, however, they do not recommend one option over others [17]. The International Patient Decision Aid Standards Collaboration (IPDAS) was founded in 2003 with the purpose of establishing an evidence-based framework for the content, development, implementation, and evaluation of patient decision aids [18]. IPDAS defines both qualifying criteria for decision aids and outcomes that can be used to evaluate the efficacy of decision aids. Commonly evaluated decisional outcomes include objective knowledge, risk perceptions, congruence of choice with patient values, decisional conflict, patient–practitioner communication, participation in the decision-making process, number of patients who reach a decision, and patient satisfaction.

A Cochrane systematic review has demonstrated that decision aids across a broad number of clinical conditions improve patients’ knowledge and accuracy of risk perceptions and improve congruency between the patients’ expressed values and their choices [19]. Patient decision aids have also been shown to decrease decisional conflict and indecision [19]. However, decision aids have not had a consistent impact on behaviors, with the exception of some leading to decreased uptake of invasive procedures or laboratory testing [19]. To our knowledge, there have not been any reviews addressing the use of patient decision aids specifically for MSK pain treatment. Accordingly, we performed a systematic review of randomized controlled trials (RCTs) evaluating the use of patient decision aids for adults making decisions about the treatment of chronic MSK pain. We were interested in evaluating the efficacy of decision aids compared with usual care and also in evaluating the relative effectiveness of decision aids to determine if a particular format, element, or approach was superior to others. Understanding which types of decision aids are most effective could inform development of future decision aids, help in the formulation of best practices for their use in clinical practice, and facilitate dissemination efforts.

Methods

Data Sources and Search Strategy

We utilized the search strategy published in the Cochrane systematic review “Decision Aids for People Facing Health Treatment or Screening Decisions” [19] but added subject headings for “chronic pain,” “back pain,” “intervertebral disc degeneration,” “osteoarthritis,” and “pain.” The complete search strategy is available in the Supplementary Data. Searches were performed of Ovid MEDLINE, Ovid MEDLINE In-Process and EPub Ahead of Print, Ovid Embase, Ovid PsycINFO, EBSCO CINAHL, and the Cochrane CENTRAL database from inception through April 2019. We also searched clinicaltrials.gov and the International Clinical Trials Registry Platform for studies that met our inclusion criteria.

Eligibility Criteria

We included published or unpublished English-language RCTs comparing the use of patient decision aids with usual care, enhanced usual care, or other decision aids in adult patients with chronic, noncancer MSK pain in the outpatient setting. We defined “decision aid” broadly and included any intervention designed to help patients make treatment decisions. Studies that evaluated a decision aid along with a co-intervention, such as motivational interviewing or an adaptive conjoint analysis (ACA) tool, were also included. Including studies that compared different formats and elements of decision aids provided us with additional data about whether certain approaches and formats were more effective for certain populations and decisions. Pilot RCT studies were included. When we identified both a pilot or feasibility study and the subsequent fully powered trial, we included the larger trial. When we identified multiple reports from the same study, we included the one that reported on the greatest number of outcomes or had the largest sample size [20]. We have referenced the other reports and their findings in the tables. We did not exclude studies based on the outcomes assessed, as we were interested in all potential results of using decision aids. We excluded studies of hypothetical decisions (e.g., if participants were not actually patients with chronic MSK pain but were asked to make a decision about treatment for a hypothetical patient with chronic MSK pain using a decision aid). Inclusion and exclusion criteria can be found in the Supplementary Data.

Study Selection

Two authors (EB and RN) independently screened all titles and abstracts identified through the search strategy against the eligibility criteria. Any abstract that was determined to be potentially relevant by at least one reviewer was selected for full-text review. Each full text was then reviewed independently by two authors (EB, RN, or NM) against the eligibility criteria, with reasons cited for exclusion. Any discrepancies were settled by discussion between all three reviewers and the senior author (UM) [20]. During this step, references were also screened for additional studies of interest.

Data Extraction

For studies that met the inclusion criteria, data were extracted independently into a customized table by the same combination of two reviewers and compared for accuracy. Data elements included population demographics, MSK condition(s) evaluated, practice setting, trial design, sample size, description and format of the patient decision aid, description of the comparator, and outcomes (see below). Any discrepancies were settled by discussion between all three reviewers and the senior author (UM) [20].

Outcomes

We evaluated outcomes related to the decision-making process and decision quality, with particular attention paid to outcomes outlined by IPDAS [18,21]. Attributes of the choice made reflect decision quality. These include objective knowledge, risk perceptions, and congruency between the chosen option and patient values. Attributes of the decision process include preparedness for decision-making, decisional conflict, satisfaction with the decision-making process, patient–provider communication, and participation in the decision-making process. Other outcomes of particular interest in MSK pain include pain intensity, function, and surgery utilization rates. Instruments used for outcome assessment are noted in Tables 2 and 3.

Table 2.

Decision-related outcomes in the intervention group compared with the control group

Reference	Attributes of the Choice	Attributes of the Decision Process	Other Decision-Related Outcomes
Studies Comparing Patient Decision Aid with Usual Care or Enhanced Usual Care
Bishop 2019 [23]	Knowledge (assessed with study-specific instrument): Greater increase in knowledge from baseline with the enhanced (PDA) website compared with the standard website	–	Informed Choice (defined as scoring above the median on the knowledge assessment and having clear and congruent attitudes and intentions on a study-specific instrument): Participants in the PDA website group were more likely to make an informed choice than those in the standard website group
Bozic 2013 [24]	–	Satisfaction with the Decision-Making Process (assessed with study-specific instrument): No statistically significant difference in patient satisfaction	More patients reached an informed decision during the first visit (defined as a knowledge score >50% on the HK-DQI [44]) Patients were more confident about what questions to ask the surgeon No statistically significant difference in stage of decision-making before appointment (assessed with SDM [45]) Lower proportion of patients undecided
de Achaval 2012 [25]	–	Decisional Conflict (assessed with DCS [46]): All groups (including control) had reductions in decisional conflict Greater reductions in the PDA alone and PDA + ACA groups The PDA alone group had the greatest improvement in the uncertainty, informed, support, and effective decision subscales of the decisional conflict score after the intervention; there was no statistically significant difference between the groups in the values clarity subscale *de Achaval 2018 [26] reported that at 12 mo postintervention, only the PDA alone group had a further decrease in decisional conflict from immediately postintervention; older age was a statistically significant predictor of higher decisional conflict	–
Fraenkel 2007 [27]	-	Preparedness for Decision-Making (assessed with PDMS [47]): ACA decision aid group was more prepared	Higher decisional self-efficacy in ACA decision aid group (assessed with Decisional Self-Efficacy Scale [48])
Ibrahim 2013 [28]	Accurate Risk Perception (assessed with Hospital for Special Surgery Knee Expectations Survey [49]): No difference Knowledge (assessed with study-specific instrument): Increase in knowledge scores from baseline in patients who received either the PDA alone or in combination with motivational interviewing No statistically significant increase in knowledge with motivational interviewing alone; no increase in knowledge in the control group	Patient–Practitioner Communication: Higher likelihood of discussing knee pain with primary care provider in all intervention groups (PDA, motivational interviewing, and PDA + motivational interviewing)	–
Patel 2014 [30]	Choice Congruent with Values (assessed with study-specific instrument): 83% in intervention and 82% in control (no statistical analysis reported)	Satisfaction with the Decision-Making Process (assessed with Satisfaction with Decision Scale [50]): No statistically significant difference in patient satisfaction	–
Stacey 2016 [31]	Accurate Risk Perception (assessed with HK-DQI [44]): No difference in estimated recovery time; higher accuracy in estimated percentage of patients who would be able to walk with less pain after surgery and estimated percentage of serious complications Good Decision Quality† (assessed with HK-DQI [44]): More patients in the intervention group achieved good decision quality, and mean knowledge scores were higher in the intervention group *Stacey 2014 [32] also reported higher decision quality and knowledge in the intervention group *Boland 2018 [33] reported a non–statistically significant difference in decision quality between sites: More patients in the intervention group achieved good decision quality at the academic clinic, while there was no difference in decision quality between groups at the community clinic	Decisional Conflict (assessed with SURE tool [51]): No statistically significant difference *Stacey 2014 [32] also reported no statistically significant difference in decisional conflict *Boland 2018 [33] reported lower decisional conflict in the intervention group at the academic clinic at 2 wk postintervention but no statistically significant difference between sites at 6 mo postintervention Preparedness for Decision-Making (assessed with PDMS [47]): More likely to know that the decision depended on their values Felt more prepared to talk to their surgeon about preferences No statistically significant difference in recognizing that a decision needs to be made or thinking about how they would like to be involved in the decision-making process *Stacey 2014 [32] reported no statistically significant difference in PDMS [47]	*Stacey 2014 [32] reported a higher proportion of undecided patients in the intervention group
Timmers 2018 [34]	Knowledge (assessed with study-specific questionnaire): Greater increase from baseline in both actual and perceived knowledge in the intervention group than the control group	Satisfaction with the Decision-Making Process (assessed with a study-specific instrument): Greater satisfaction with knowledge and information provided in the intervention group; no difference in overall satisfaction with the consultation between the 2 groups	No difference in patient perception of shared decision-making between the 2 groups Greater confidence about treatment choice in the intervention group
Veroff 2011 [35]	Knowledge (assessed with study-specific instrument): Higher knowledge scores in both the PDA alone group and the PDA + health coaching group	Decisional Conflict (assessed with study-specific instrument): No statistically significant difference	–
Wilkens 2019 [37]	–	Decisional Conflict (assessed with DCS [46] immediately after intervention): Lower decisional conflict in the PDA group Satisfaction with the Decision-Making Process (assessed with study-specific instrument immediately after the intervention): No statistically significant difference in satisfaction with the visit	No statistically significant difference in decision regret (assessed with Decision Regret Scale [52]) at 6 wk or 6 mo postintervention
Comparisons Between Patient Decision Aids
Allen 2016 [38]	Knowledge (assessed with HK-DQI [44] knowledge subscale): Both the Internet and the DVD decision aid groups showed a significant increase in knowledge from baseline over time, and there was no statistically significant difference between the 2 groups	Decisional Conflict (assessed with DCS [46]): Both the Internet and DVD decision aid groups showed a significant decrease in decisional conflict over time, and there was no statistically significant difference between the 2 groups Preparedness for Decision-Making (assessed with PDMS [47]): There were higher PDMS scores in the DVD group	Higher stage of decision-making achieved in the DVD group (assessed with SDM [45])
Deyo 2000 [39]	*Phelan 2001 [40] evaluated knowledge (assessed with a study-specific instrument) in a subset of 100 early participants in this trial and reported a greater increase in knowledge score from baseline in the combined video and booklet group than in the booklet alone group; the improvement was greatest in those with the lowest baseline knowledge scores	Satisfaction with the Decision-Making Process (assessed with study-specific instrument): No difference	No difference in satisfaction with medical care
Mangla 2019 [41]	Knowledge (assessed with HK-DQI [44]): There was no statistically significant difference in knowledge scores between the 2 decision aid groups; knowledge scores were higher for patients who reviewed all of PDA-B compared with patients who reviewed only part of PDA-B; there was no statistically significant difference in knowledge scores according to the amount of PDA-A reviewed; knowledge scores also did not differ by literacy or age	-	There was no statistically significant difference in patient goals and concerns (assessed with HK-DQI [44]) between the 2 decision aid groups
Shue 2016 [42]	Knowledge (assessed with study-specific instrument): There was no statistically significant increase in knowledge score from baseline in either the booklet alone or the booklet + DVD group	Participation in Decision-Making (assessed with study-specific instrument): No statistically significant difference Satisfaction with the Decision-Making Process (assessed with study-specific instrument): No statistically significant difference between groups in satisfaction with knowledge gained in the decision process	No statistically significant difference in stage of decision-making (assessed with study-specific instrument) No statistically significant difference in number of patients who changed their treatment preference after the decision aid
Weymann 2015 [43]	Knowledge (assessed with study-specific instrument): Higher knowledge scores with the tailored app than the nontailored app	Decisional Conflict (assessed with DCS [46]): No statistically significant difference between groups Preparedness for Decision-Making (assessed with PDMS [47]): No statistically significant difference between groups	Patient Empowerment (assessed with Health Education Impact Questionnaire [53]): No statistically significant difference between groups

ACA = adaptive conjoint analysis; DCS = Decisional Conflict Scale; HK-DQI = Hip/Knee Osteoarthritis Decision Quality Instrument; PDA = patient decision aid; PDMS = Preparation for Decision-Making Scale; SDM = Stage of Decision-Making Scale.

*As described in the text, these are references using the same data set. Relevant findings are included in the table.

^†Combined both “Knowledge” and “Chosen Option Congruent with Values” outcomes into “Good Decision Quality,” which was defined as both a score >66% on the knowledge test and a choice that corresponded with stated values.

Table 3.

Other musculoskeletal pain-related outcomes in the intervention group compared with the control group

Reference	Surgery-Related Outcomes	Pain/Function-Related Outcomes	Other Outcomes
Studies Comparing Patient Decision Aid with Usual Care or Enhanced Usual Care
Bishop 2019 [23]	–	–	No differences in beliefs about acupuncture or willingness to try acupuncture between the 2 groups (assessed with the Low Back Pain Treatment Beliefs Questionnaire [54])
Bozic 2013 [24]	No statistically significant difference in percentage of patients choosing surgery	–	Surgeons felt that patients asked a greater number of questions and more appropriate questions Surgeons felt that the visits were more efficient Surgeons were more satisfied with the consultation No statistically significant difference in length of consultation or amount of face-to-face time with the surgeon
Fraenkel 2007 [27]	–	–	Higher arthritis self-efficacy scores in the ACA decision aid group (assessed with the Arthritis Self-Efficacy Scale [55]) The majority of patients felt the ACA decision aid task was easy to complete, accurately reflected their preferences, and would recommend it to others with knee pain (assessed with study-specific instrument)
Ibrahim 2013 [28]	No statistically significant difference in number of referrals to orthopedics No statistically significant difference in attendance of orthopedic appointments	–	One month after the intervention, all groups including the control were more willing to consider TKR; there was no statistically significant difference between the groups 3 mo after the intervention, only the PDA alone group sustained the increase in willingness to consider TKR The increase in willingness to consider TKR in the combined PDA and motivational interviewing group was not statistically significant at any time point (all assessed with the Willingness Rating [56])
Ibrahim 2017 [29]	Higher rate of TKR at 12 mo No statistically significant difference in receipt of recommendation for surgery from orthopedic surgeon within 6 mo	–	–
Patel 2014 [30]	–	Less improvement in the Roland Morris Disability Questionnaire [57] No statistically significant difference in Modified Von Korff disability score [58], Modified Von Korff pain score [58], physical component of SF-12 [59], mental component of SF-12 [59], anxiety and depression (assessed with the Hospital Anxiety and Depression Scale [60]), pain self-efficacy (assessed with the Pain Self-Efficacy Questionnaire [61]), fear avoidance beliefs (assessed with the Fear Avoidance Beliefs Questionnaire [62])	No statistically significant difference in satisfaction with treatment (assessed with study-specific instrument) Incremental cost-effectiveness ratio of £1900 per quality-adjusted life-year gained
Stacey 2016 [31]	No statistically significant difference in percentage of patients choosing surgery Non–statistically significant trend toward shorter wait times for surgery in the intervention group *Stacey 2014 [32] reported no statistically significant difference in percentage of patients choosing surgery and no difference in wait times for surgery *Boland 2018 [33] reported no difference in surgery rates or wait times between sites	–	–
Timmers 2018 [34]	No difference between groups in the choice for operative treatment	–	Patients in the control group were less sure if their symptoms were caused by OA (assessed with study-specific instrument)
Veroff 2011 [35]	No statistically significant difference in rate of surgery	–	No statistically significant difference in total medical costs per member per month No statistically significant difference in medical costs that can be impacted per member per month
Vina 2016 [36]	No statistically significant difference in percentage of patients referred to an orthopedic surgeon within 12 mo of intervention	–	In patients recruited from a university clinic, there was increased willingness to consider TKR at 3 mo postintervention; otherwise, there was no statistically significant difference in willingness to consider TKR at any time point (assessed with a previously evaluated 5-point ordinal response scale [63])
Wilkens 2019 [37]	No statistically significant difference in number of patients choosing surgical vs nonsurgical treatment	No statistically significant difference in pain (assessed with a study-specific scale) immediately after the visit, at 6 wk postintervention, or at 6 mo postintervention No statistically significant difference in upper extremity disability (assessed with the short form of the Disabilities of the Arm, Shoulder, and Hand questionnaire (QuickDASH questionnaire) [64]) immediately postintervention	No statistically significant difference in depression (assessed with the Patient Health Questionnaire-2 [65]) immediately postintervention No statistically significant difference in physician empathy (assessed with the Consultation and Relational Empathy Measure (CARE) [66]) immediately postintervention No statistically significant difference in duration of consultation No statistically significant difference in satisfaction with treatment at 6 wk or 6 mo postintervention (assessed with study-specific instrument) No statistically significant difference in the percentage of patients who changed their treatment choice or their surgeon at 6 wk or 6 mo postintervention
Comparisons Between Patient Decision Aids
Allen 2016 [38]	–	–	More patients in the Internet group thought the length of the decision aid was “just right” More patients in the Internet group thought there was too little information More patients in the DVD group thought the decision aid included enough information to help them decide on treatment More patients in the DVD group thought the decision aid would be helpful when making a decision about OA treatment (all assessed with the Acceptability of Decision Aids Scale [67])
Deyo 2000 [39]	No statistically significant difference in percentage of total patients choosing surgery *Phelan 2001 [40] also reported no difference in preference for surgery among a subset of 100 early participants Fewer patients in the herniated disk subgroup chose surgery	No statistically significant difference in Roland Morris Disability Score [57] No statistically significant difference in back or leg pain severity No statistically significant difference in satisfaction with symptoms (assessed with a previously validated questionnaire [68])	No statistically significant difference in employment status No statistically significant difference in percentage of patients seeking or receiving compensation More patients in the video and paper PDA group felt they received adequate information about their back condition No statistically significant difference in percentage of patients who felt they received adequate information about treatment, had the desired amount of input in choice of treatment, felt their opinion was important in the treatment decision, relied too much on doctors’ opinions, or felt they made the right choice No statistically significant difference in patient rating of explanations of medical procedures and tests, personal interest expressed by doctors and nurses, reassurance and support offered by doctors and nurses, and amount of time spent with doctors and nurses (all assessed with study-specific instruments)
Mangla 2019 [41]	No statistically significant difference in preference for surgery between the 2 decision aid groups No statistically significant difference in the percentage of patients who received their preferred treatment between the 2 groups	–	More patients in the PDA-A group reviewed the entire decision aid than did patients in the PDA-B group (verified statistical significance with author)
Shue 2016 [42]	–	–	More patients in the booklet and DVD group felt the decision aid should be shorter

ACA = Adaptive Conjoint Analysis; OA = osteoarthritis; PDA = Patient Decision Aid; SF-12 = 12 Item Short Form Health Survey; TKR = total knee replacement.

*As described in the text, these are references using the same data set. Relevant findings are included in the table.

Risk of Bias

Risk of bias was assessed for each retained study by two separate reviewers (EB, RN, or NM) and confirmed by the senior author (UM) using the Cochrane Risk of Bias Tool [22]. Any discrepancies were resolved by in-person discussion until there was complete agreement [20]. We assessed the risk of bias as high, low, or unclear in each category: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other bias. Due to the nature of patient decision aids as an intervention, it is not possible to perform a truly blinded trial, because patients will know that they received an intervention designed to help them make a decision. We assessed the risk of bias for participant blinding as low if there was an active control arm, such as an informational booklet. Blinding of outcome assessment is more important for trials of decision aids, particularly when assessing subjective outcomes. If clarification was needed to make a risk of bias assessment, we e-mailed the study’s corresponding author and incorporated any additional information provided into our assessment.

Results

We identified 526 records using the search strategy from a Cochrane Systematic Review [19], four through hand-searching references, eight from clinicaltrials.gov, and nine from the International Clinical Trials Registry Platform. Two reviewers (EB and RN) reviewed a total of 477 abstracts after 70 duplicates were removed. After abstract review, 64 articles were selected for full-text review. After full-text review, 17 studies met our inclusion criteria (listed in Table 1). The search and article evaluation process is outlined in Figure 1.

Table 1.

Characteristics of included studies

Reference	Chronic Pain Condition	Participants	Decision	Intervention	Comparator
Studies Comparing Patient Decision Aid with Usual Care or Enhanced Usual Care
Bishop 2019 [23]	Back pain	N = 350; 175 enhanced website (PDA) and 175 standard website Mean age 48 years 43.7% male and 56.3% female	Acupuncture vs no acupuncture	“Enhanced” website designed to increase knowledge, support informed decision-making, and promote realistic expectations among patients considering acupuncture	“Standard” website that provides brief information about acupuncture, its safety, possible side effects, and contraindications
Bozic 2013 [24]	Hip OA or knee OA	N = 123; 61 intervention and 62 control Mean age 63 years 45.5% male and 54.5% female	Joint replacement vs medical management	Video and booklet PDA produced by the Informed Medical Decisions Foundation and Health Dialog; in addition, patients received a preconsultation telephone call with a health coach to construct a list of questions for the surgeon; after the consultation, patients were mailed an audio-recording of the consultation and a copy of the surgeon’s note	Handouts with information about the clinic and also the signs and symptoms, diagnosis, and treatment options for hip and knee OA; patients received a telephone call to remind them about the appointment and confirm receipt of materials
de Achaval 2012 [25]	Knee OA	N = 208; 70 PDA alone, 69 PDA + ACA software, 69 control Mean age 63 years 32% male and 68% female	Joint replacement vs medical management	Video and booklet PDA produced by the Foundation for Informed Medical Decision Making and Health Dialog or a combination of the PDA and a computer-based ACA tool; the ACA tool was designed to help patients clarify their values related to the decision	Patient education booklet about OA from NIAMS
	*de Achaval 2018 [26] reported on decisional conflict 12 mo postintervention in 189 of these participants.
Fraenkel 2007 [27]	Knee OA	N = 83; 43 ACA PDA, 40 control Mean age 74 years % male and female not reported	Treatment options: capsaicin vs acetaminophen vs anti-inflammatory drugs vs intra-articular injections vs exercise vs combination of exercise and medications	Computer-based ACA PDA designed to help patients clarify their preferences and values related to the treatment options; patients received a printed summary of their preferences	Patient education booklet from the Arthritis Foundation
Ibrahim 2013 [28]	Knee OA	N = 639; 162 PDA alone, 158 motivational interviewing alone, 158 PDA + motivational interviewing, 161 control Mean age 61 years 94% male and 6% female	Joint replacement vs medical management	Video PDA produced by the Foundation for Informed Medical Decision-Making or motivational interviewing or a combination of the decision aid video and motivational interviewing	Patient education booklet about OA from NIAMS
Ibrahim 2017 [29]	Knee OA	N = 336; 168 intervention and 168 control Mean age 59 years 30% male and 70% female	Joint replacement vs medical management	Video PDA produced by the Informed Medical Decisions Foundation	Patient education booklet about OA from NIAMS
Patel 2014 [30]	Back pain	N = 148; 85 intervention and 63 control Mean age 48 years 33.8% male and 66.2% female	Treatment options: structured group exercise vs manual therapy vs acupuncture vs cognitive behavioral therapy vs self-management	Booklet PDA created by researchers and clinicians in combination with shared decision-making training for physiotherapists	Usual care (no decision support information or shared decision-making training for physiotherapists)
Stacey 2016 [31]	Hip OA or knee OA	N = 334; 167 intervention and 167 control Mean age 66 years 42.5% male and 57.5% female	Joint replacement vs medical management	Video and booklet PDA produced by the Informed Medical Decisions Foundation; a report summarizing the patient’s preferences and clinical assessment was also provided to the surgeon	Information about joint replacement surgery (without decision-making support for the patient) and a clinical assessment report for the surgeon
	*Stacey 2014 [32] is a pilot RCT for Stacey 2016 that was conducted in patients with knee OA. *Boland 2018 [33] is a subgroup analysis of Stacey 2016 comparing outcomes in patients from an academic clinic with those in a community clinic.
Timmers 2018 [34]	Knee OA	N = 213; 91 intervention and 122 control Mean age 62 years 47.4% male and 52.6% female	Surgical vs medical management	Interactive PDA app for smartphone or tablet called the Patient Journey App	Standard education consisting of at least a website and an informational event
Veroff 2011 [35]	Knee OA or hip OA or back pain or chronic knee pain	N = 9,925; 3,311 PDA only, 3,323 PDA with health coaching, and 3,291 controls Mean age 56 years 43% male and 57% female	Surgical vs medical management	Video and booklet PDA or a combination of the PDA and health coaching over the telephone	No support
Vina 2016 [36]	Knee OA	N = 490; 238 intervention and 252 control Mean age 62 years 49% male and 51% female	Joint replacement vs medical management	Video PDA produced by the Foundation for Informed Medical Decision-Making and motivational interviewing	Patient education booklet about OA from NIAMS
Wilkens 2019 [37]	Trapeziometacarpal OA	N = 90; 45 PDA and 45 control Mean age 63 years 27.8% male and 72.2% female	Surgical vs nonsurgical management (options include observation, pain medication, injection, and orthosis)	Interactive, Internet-based PDA	Usual care, including an informational brochure from the American Society of Surgery of the Hand
Comparisons Between Patient Decision Aids
Allen 2016 [38]	Hip or knee OA	N = 155; 75 Internet PDA and 80 DVD PDA Mean age 62 years 39.4% male and 60.6% female	Joint replacement vs medical management	Video PDA produced by the Informed Medical Decisions Foundation and Health Dialog with an accompanying booklet	Internet-based PDA produced by the Informed Medical Decisions Foundation and Health Dialog
Deyo 2000 [39]	Back pain	N = 393; 190 PDA, 203 control Mean age 52 years 52% male and 48% female	Surgical vs medical management	Video PDA plus a paper PDA booklet	Paper PDA booklet alone
*Phelan 2001 [40] reported knowledge and preference for surgery on a subset of 100 early participants in this trial.
Mangla 2019 [41]	Hip OA or knee OA	N = 58; 33 PDA-A and 25 PDA-B Mean age 64 years 48% male and 52% female	Joint replacement vs medical management	PDA-A: 15-page printed brochure PDA from Healthwise	PDA-B: Combination video and booklet PDA from Health Dialog
Shue 2016 [42]	Hip or knee OA	N = 132; 66 in the booklet plus DVD group and 66 in the booklet alone group Mean age 61 years 47% male and 53% female	Joint replacement vs medical management	Booklet and DVD PDA with patient testimonials and physician interviews	Paper PDA booklet
Weymann 2015 [43]	Back pain	N = 378; 190 in the tailored delivery arm and 188 in the standard delivery arm Mean age 51 years 35.5% male and 64.5% female (author provided demographics specific to the patients with chronic low back pain)	Treatment of chronic low back pain, including both pharmacologic and nonpharmacologic options	Internet-based interactive health communication application with information about chronic low back pain, its diagnosis, and treatment options; the intervention “tailored” content delivery to the patient’s coping style, baseline knowledge, and preference for level of detail	“Nontailored” Internet-based interactive health communication application with information about chronic low back pain, its diagnosis, and treatment options
*Note that this study also evaluated a decision aid for patients with Type 2 Diabetes. We have reported only the data on patients with chronic low back pain.

ACA = adaptive conjoint analysis; OA = osteoarthritis; PDA = patient decision aid; RCT = randomized controlled trial; NIAMS = National Institute of Arthritis and Musculoskeletal and Skin Diseases.

* As described in the text, these are references using the same data set. Relevant findings are included in the table.

Figure 1.

Flowchart of the search and article evaluation process.

Characteristics of Study Participants

The total number of patients included in all the trials was 14,055. The trial sizes ranged from 58 to 9,925 participants. The patients in the included trials were on average 57 years old, while the mean sample age across the studies ranged from 48 to 74 years. With the exception of Fraenkel [27], which only included adults who were at least 60 years old, none of the trials focused exclusively on older adults. Across all the trials, 55% were women and 45% were men. The means are heavily influenced by Veroff [35], which was the largest trial and included 9,925 participants. Six studies [25,27–29,34,36] evaluated only patients with knee OA. Five studies [24,31,38,41,42] evaluated patients with either hip or knee OA. Four studies [23,30,39,43] evaluated patients with chronic back pain. One study [35] evaluated patients with knee pain, back pain, knee OA, and/or hip OA. One study [37] evaluated patients with trapeziometacarpal (TMC) OA.

Types of Decisions

In 13 studies [24,25,28,29,31,34–39,41,42], patients were deciding between surgical and conservative management of chronic pain conditions, including joint replacement and back surgery. In the other four studies [23,27,30,43], patients used decision aids to decide among nonsurgical treatment options, including acupuncture, medications, injections, exercise-based therapy, and cognitive behavioral therapy.

Characteristics of the Decision Aids

Eleven [24,25,28,29,31,35,36,38,39,41,42] of the decision aids studied included videos and were studied either alone or in combination with additional interventions such as motivational interviewing, ACA software, or patient preference reports for the surgeon. Allen [38], Bishop [23], and Wilkens [37] used Internet-based patient decision aids. Timmers [34] and Weymann [43] used interactive patient decision aid apps. Fraenkel [27] used a computer-based ACA decision aid. Patel [30] used a paper decision aid in combination with shared decision-making training for physiotherapists. Twelve studies [23–25,27–31,34–37] compared the decision aids with usual care. Allen [38] compared a video decision aid to an Internet-based decision aid. Deyo [39], Mangla [41], and Shue [42] compared a combined video and booklet decision aid with a booklet decision aid alone. Weymann [43] compared an app tailored to the user’s coping style, baseline knowledge, and preference for level of detail with a nontailored app. Three studies [28,29,36] were designed to evaluate the role of patient decision aids in reducing the disparity in rates of knee replacement surgery in black (vs white) patients with OA.

Decision-Related Outcomes

Decision-related outcomes are reported in Table 2 and are grouped by the IPDAS outcomes for evaluating patient decision aids, attributes of the choice made, and attributes of the decision process. Additionally, studies that compared decision aids with usual care are separated from studies making comparisons between decision aids.

Attributes of the Choice Made (Decision Aids vs Usual Care)

Five studies [23,28,31,34,35] evaluated the patients’ knowledge about the disease and treatment options, and all five found an improvement in knowledge with the use of a patient decision aid. Stacey [31] evaluated “decision quality,” which was a combined outcome of knowledge and a choice that corresponded with the patient’s stated values. Higher decision quality was achieved in the decision aid group. Two studies [28,31] evaluated the accuracy of risk perception in patients considering total knee replacement. Ibrahim [28] found no difference in the Hospital for Special Surgery Knee Expectations Scale [49] scores in the group that used the video decision aid as compared with usual care. Stacey [31] found that patients who used the decision aid more accurately predicted the percentage of serious complications and the percentage of patients who would be able to walk with less pain after surgery but had no improvement in accuracy of estimating recovery time compared with the group who received usual care.

Attributes of the Decision Process (Decision Aids vs Usual Care)

Four studies [25,31,35,37] evaluated decisional conflict. de Achaval [25] and Wilkens [37] found lower decisional conflict with use of the decision aid. In de Achaval’s study [25], the group that received only the decision aid had a greater reduction in decisional conflict than the group that received the decision aid in combination with the ACA software. Stacey [31] and Veroff [35] found no statistically significant difference in decisional conflict with use of the decision aid.

Two studies [27,31] evaluated how prepared patients felt to make a decision. Stacey [31] found mixed results on the Preparation for Decision-Making Scale [47] items. There was no statistically significant difference in patients recognizing that a decision needs to be made or thinking about how they would like to be involved in the decision-making process. However, patients who received the decision aid were more likely to know that the decision depended on their values and felt more prepared to talk to their surgeon about their preferences. Fraenkel [27] demonstrated higher preparedness for decision-making with use of the ACA decision aid.

Four studies [24,30,34,37] evaluated satisfaction with the decision-making process and found no difference in overall patient satisfaction with use of a decision aid compared with usual care, although Timmers [34] did find that use of the decision aid increased satisfaction with the information provided.

Ibrahim [28] evaluated patient–practitioner communication and found a higher likelihood of discussing knee pain with the primary care provider when a decision aid was used.

Surgery-Related Outcomes (Decision Aids vs Usual Care)

Eight studies [24,28,29,31,34–37] evaluated outcomes related to surgery, including patient preference for surgery, wait times for surgery, and referrals to orthopedics. Results are reported in Table 3, with studies that compared decision aids with usual care separated from studies making comparisons between decision aids. Only Ibrahim [29] found that patients using the decision aid had a higher rate of total knee replacement (TKR) at 12 months compared with controls. The rest of the studies found no statistically significant difference in outcomes related to surgery with use of a decision aid.

Pain/Function Outcomes (Decision Aids vs Usual Care)

Patel [30] and Wilkens [37] evaluated outcomes related to pain and function (Table 3). Patel [30] used a paper booklet to help patients choose between treatment strategies employed by physiotherapists for chronic lower back pain, including group exercise, manual therapy, acupuncture, cognitive behavioral therapy, and self-management. The therapists in the intervention group received training in shared decision-making. There was no statistically significant difference between the intervention and control groups in patient satisfaction with treatment outcome, Modified Von Korff disability or pain scores [58], the physical or mental components of the 12-Item Short Form Health Survey [59], anxiety, depression, pain self-efficacy, or fear avoidance beliefs. There was less improvement in the Roland Morris Disability Questionnaire scores [57] in the intervention group as compared with the control group. Wilkens [37] did not find a difference in upper extremity disability between the group that used the decision aid and the usual care group.

Comparisons Between Decision Aids

Five studies made comparisons between decision aids [38,39,41–43]. Three [39,41,42] compared a paper decision aid with a combined paper and video decision aid. Deyo [39] found greater improvement in knowledge with the combined paper and video decision aid, but Mangla [41] and Shue [42] found no difference in knowledge. There was also no difference in satisfaction with the decision-making process or participation in decision-making between the two types of decision aids. Further, there was no difference in pain and function [39] or surgery-related outcomes [39,41] between the two types of decision aids.

Allen [38] compared a video decision aid with an Internet-based decision aid and found that patients who used the video decision aid were more prepared for decision-making and achieved a higher stage of decision-making than those who used the Internet-based decision aid, but otherwise there was no statistically significant difference in outcomes.

Weymann [43] compared a decision aid app tailored to the user’s coping style, baseline knowledge, and preference for level of detail with a nontailored app and found higher knowledge scores with the tailored app but no difference in decisional conflict, preparedness for decision-making, or patient empowerment.

The addition of motivational interviewing [28], an ACA tool [25], or telephone-based health coaching [35] to a decision aid did not improve any of the outcomes compared with the use of the decision aid alone. Additional outcomes of interest are reported in Table 3.

Risk of Bias

The risk of bias for each study, along with the full rationale for each decision, is provided in the Supplementary Data. Across all the studies, there was only one high risk of bias rating in a single category.

Discussion

In this systematic review, we found 17 RCTs evaluating the use of patient decision aids for adults making treatment decisions regarding chronic MSK pain. Thirteen of the 17 studies evaluated patient decision aids designed to help patients choose between surgical versus medical management of chronic MSK pain. The effects of the decision aids on decision-related outcomes in our review were mixed. All studies that assessed participant knowledge found improvement in this outcome. None of the studies that assessed satisfaction with the overall decision-making process found improvement. Inconsistent results were found regarding the impact of decision aids on other decision-related outcomes.

Our results are in contrast with the findings reported in the 2017 Cochrane Review [19]. This review evaluated multiple decision aids used in a wide variety of health care decisions and concluded that decision aids improve patients’ knowledge, accuracy of risk perceptions, congruence between expressed values and choices, decisional conflict, and indecision. This discrepancy raises the possibility that decisions regarding the treatment of chronic MSK pain may be less suitable to the use of patient decision aids than other health care decisions. In contrast to many other medical decisions, the treatment of chronic MSK pain often involves multiple modalities simultaneously and thus may require a unique decision-making approach. Additionally, the mixed results of patient decision aids on decisional conflict suggests that the provision of information alone may not be enough to help patients make an informed decision. Rather, patient decision aids are only one part of the shared decision-making process, and the provider plays a key role in helping patients synthesize information so they make the most informed, appropriate decision in the context of their own values and goals.

Most of the trials we identified evaluated the use of video decision aids. Four compared the videos with either paper or Internet-based decision aids, and the evidence did not suggest that one particular format was clearly more effective. Some trials showed that augmenting the patient decision aids with other interventions such as motivational interviewing and health coaching did not improve efficacy. In fact, de Achaval [25] found that patients who received the decision aid alone without the ACA software tool scored better on certain subscales of the decisional conflict score. However, in a separate trial, when the decision aid consisted of solely an ACA tool, it effectively increased preparedness for decision-making and decisional self-efficacy [27]. One hypothesis is that too many interventions or interventions that take too long to complete may actually increase a patient’s uncertainty with a decision and dilute the effect of the decision aid.

The majority of trials evaluated in the current study focused on decisions related to surgery vs nonsurgical management. This restricts the generalizability of our findings to the broader chronic MSK pain population that may not be interested in surgical intervention and highlights an important knowledge gap in the existing literature. Knee and hip replacement and hand surgery are important treatment options for patients with advanced OA refractory to medical management [69,70]. However, the decision to undergo elective surgery is a highly preference-sensitive decision and requires patient commitment to the rehabilitation process afterward to achieve optimal outcomes. This is an ideal scenario for the use of a patient decision aid and is likely the reason that most trials analyzed in the current study focused on this topic. Most of the studies that evaluated the effect of the decision aid on the decision to undergo surgery found no difference. This is in contrast to the results presented in the 2017 Cochrane Review [19] and observational data [71] that demonstrated that use of decision aids reduced the number of people choosing elective surgery over conservative management. This may be because many of the patients considering knee or hip replacement have such advanced disease that there is no other reasonable management option to maintain a certain quality of life. Though there is not robust evidence that use of a decision aid improves pain and function, more distal clinical outcomes to the decision-making process, it seems reasonable to use the available patient decision aids to help patients decide between surgical and conservative management of OA (knee, hip, and TMC) and chronic back pain with the goal of improving the decision-making process and decision quality.

For many patients with chronic MSK pain, surgery is not the ideal intervention, certainly not firstline management. Patients who choose not to undergo joint replacement or back surgery still face preference-sensitive decisions about the medical and rehabilitative management of their chronic MSK pain. These decisions can be complex, especially in older adults, because the etiology of chronic MSK pain is often multifactorial, with contributions from anatomic, physiologic, and psychological factors [72,73]. Additionally, older adults often have multiple other chronic medical conditions and take medications for these conditions that frequently limit the pharmacologic treatment options for chronic MSK pain. There are several evidence-based, nonpharmacologic approaches to chronic pain that are implementation-ready in certain populations, including cognitive behavioral therapy, mindfulness-based therapy, exercise therapy, and manipulation [15,74,75]. We found only four trials of decision aids [23,27,30,43] designed to help patients choose between conservative management strategies for chronic MSK pain. Yet, there are many preference-sensitive decisions about the treatment of chronic MSK pain for which the use of decision aids might be appropriate, such as deciding among medications and their respective side effect profiles, deciding whether to start or stop opiates, and deciding between behavioral, psychological, and rehabilitative therapies (especially when resources are limited).

A limitation of our review is the many studies with unclear risk of bias in at least one category due to incomplete information. Additionally, although we performed a comprehensive search for MSK pain (that included OA and chronic back pain), we may have missed other chronic pain conditions that have multiple reasonable treatment options for which a patient decision aid would be appropriate.

Another notable limitation identified by this study is the lack of research focused on older adults. We originally intended to focus our search on older adults, but we had to broaden our strategy to include trials of all adults when we did not find any studies that limited the patient population to adults over the age of 65. The average age of the patients from this review was only 57 years, but we know that the prevalence of chronic MSK pain and resulting functional limitation increase with age [1]. A systematic review of decision aids designed to help older adults make various health decisions found that these tools improved knowledge, led to more accurate risk perception, reduced decisional conflict, and enhanced patient participation [76]. Future research on patient decision aids for MSK pain should consider whether they are equally effective in younger and older populations and why discrepancies exist if they do. As the priorities and goals of care often change in later life, older adults will especially benefit from tools designed to help them clarify their values and make health care decisions that accurately reflect those values [77].

Existing research suggests that older adults may not be well informed about the treatment options for chronic MSK pain and that, when informed, their preferences may differ from providers’ practice patterns [6–8,11,12]. Because older adults value the patient–provider relationship and information from their providers [5,9,10], patient decision aids present an opportunity to provide accurate information about treatment options, elicit preferences, and engage both the provider and the older patient in shared decision-making to create a treatment plan that aligns with patient values. Especially for older adults, who often face multimorbidity and polypharmacy and may not be surgical candidates, patient decision aids focused on nonpharmacologic and nonsurgical approaches (the recommended firstline therapies) are needed [78–81]. As chronic MSK pain is often multifactorial in etiology, future decision aids should also include options for multimodal treatment approaches. Because of age-related decline in deliberative decision-making capacity with relative preservation of affective decision-making capacity [82], older adults may make decisions differently than younger adults, and thus may benefit from decision aids that account for these age-related changes.

In summary, our review of patient decision aids for chronic MSK pain suggests that the use of patient decision aids may improve knowledge for adults deciding whether to pursue surgery. However, the inconsistent effects on other decision-related outcomes imply that patient decision aids alone may not be sufficient to ensure high-quality shared decision-making about treatment for chronic MSK pain. Future research should investigate the role decision aids can play in a comprehensive approach to shared decision-making for chronic MSK pain that also focuses on the role of the provider. Our work also highlights the need for the development of decision aids focused on nonsurgical options in the management of chronic MSK pain, including medications, nonpharmacologic treatment, and behavioral interventions [81]. Future development and evaluation of decision aids for chronic MSK pain management should consider other technology-based delivery systems, expanding the scope of modalities offered to include psychological and behavioral therapies, targeting patients with varying levels of health literacy, and focusing on the large population of older adults with MSK pain.

Supplementary Data

Supplementary data are available at Pain Medicine online.

Funding sources: Dr. Makris is a VA HSR&D Career Development awardee at the Dallas VA (IK2HX001916) and supported in part (along with Mrs. Mayo and Dr. Halm) by a grant from the Agency for Healthcare Research and Quality (R24 HS022418) at UT Southwestern Medical Center. Dr. Reid is supported by an Edward R. Roybal Translational Research on Aging award (P30AG022845) from the National Institute on Aging. Dr. Reid is also supported by a National Institute on Aging award (K24AGO53462), an investigator-initiated award from Pfizer Pharmaceuticals, and the Howard and Phyllis Schwartz Philanthropic Fund. Dr. Fraenkel is supported by the National Institute of Arthritis and Musculoskeletal and Skin Diseases, part of the National Institutes of Health, under Award Number AR060231-06. Dr. Weiner is supported by the National Institute of Drug Abuse under BAA-N01DA-15-4422 (Center of Excellence in Pain Education).

Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Conflicts of interest: EB, RN, NM, HM, LF, DW, EH, and UEM have no conflicts of interest to disclose. In the past two years, Dr. Reid has served as a consultant for Johnson and Johnson.