Introduction

Osteoarthritis (OA) of the knee is a common degenerative disorder characterized by damage to articular cartilage [1–4]. While no cure exists for knee osteoarthritis (OA), treatment algorithms for OA include minimally invasive therapies such as physical therapy, pain medication, bracing, corticosteroid, and hyaluronic acid (HA) injections. The goal of nonoperative management is to preserve quality of life, improve mobility, and reduce pain [5–8]. When patients struggle with activities of daily living while following non-operative management, surgical intervention may be considered to restore patient mobility and reduce pain. The decision to pursue total knee arthroplasty (TKA) requires the patient to be well informed and consider a multitude of factors. This poses a challenge as patients come from a variety of backgrounds with different levels of health literacy regarding treatment options. Moreover, baseline patient health literacy can be difficult to assess during the surgical consultation [9–12]. Shared-Decision Making Tools (SDMTs) seek to address this health literacy disparity.

Patient decision aids are one solution to improve patient comprehension through easily digestible terminology, images, and animations [13–16]. They provide a supplemental source of information to educate patients about their condition and consequences of different treatment regimens. Despite their promise, some decision aids reported in the literature have failed to demonstrate improvement in the patient’s baseline knowledge over the standard of care orthopaedic surgery consultation [17–19]. Additionally, many decision-making aids (reading materials, videos, etc.) previously studied were not personalized to a patient’s individual circumstances, were administered by research staff rather than clinicians, or were not administered within a surgical clinical setting [17, 19–24]. There is a paucity of data reflecting the effectiveness of a decision-making tool that is customized for each patient, by adapting to inputs from both the patient and the clinician during the clinic visit.

This study aimed to evaluate the use of a patient specific SDMT in a population of patients receiving treatment for knee OA. The tool was designed to mitigate disparities in health outcomes in knee OA patients through a collaboration between the non-profit organization, Movement is Life, and the Johns Hopkins School of Public Health [25]. The tool is first-of-its-kind in that it incorporates demographic information, health insurance status, past medical history, and biometric data to make personalized predictions of a patient’s expected outcomes over a 1-, 3-, and 6-year period [24]. A version of this SDMT was reported in the context of treating historically marginalized patient populations within a primary care practice setting supported by a clinical research coordinator [24]. A previous randomized controlled study with the SDMT demonstrated increased self-reported physical activity at one month [24]. Based on these encouraging results, Movement is Life developed an interactive, online tool available for broader use that could be utilized during the orthopaedic clinic visit.

Given the success of the tool within the primary care setting, and with the understanding that most patients decide whether to pursue surgery within a surgical clinical setting, the SDMT was instituted at an adult reconstruction surgery clinic. In the present study, patient visits using the SDMT were executed by orthopaedic surgeons and advanced practice providers (APPs). This study aimed to measure the effectiveness of the SDMT as used by the orthopaedic surgeons and APPs in patients with knee OA.

Methods

Shared Decision-Making Tool

This SDMT has been described previously in the literature (Bansback et al,. 2022; Barry, 2022; Bossen et al., 2022). A Markov model served as the basis for the SDMT, allowing predictive modeling based on data from the current functional state of the patient [26]. The Kellgren-Lawrence scale was used to measure disease progression, and age, sex, race/ethnicity, and comorbidity load were included as model parameters [25, 27]. Using published guidelines, private payer claims data (Truven Health Analytics, IBM Watson), and clinical experts, ten clinically appropriate treatment pathways were identified, and their effectiveness was evaluated using data from the literature [25].

Patient and clinician focus groups found that typical health economic outputs, such as Quality Adjusted Life Years (QALY) and Incremental Cost-Effectiveness ratios (ICERS) were too specialized for the average patient to understand. The focus groups identified the treatment options most impactful to patient outcomes related to improved quality of life: likely future pain, likely future activity ability, and future lost income [28]. The model estimated both direct medical costs, based on Medicare reimbursement rates, and an assumption of 133% of Medicare rates for private insurers and indirect medical costs based on expected lost wages to the patient related to missed work. The financial burden of the disease from missing employment was stratified by pain and educational level [25]. A simple, one-page output of the tool was developed for patient consumption (Fig. 1).

The tool is unique in that it incorporates race, education level, income/occupation, health insurance status and type, past medical history, and biometric data to make personalized predictions for each patient’s expected outcomes over a 1-, 3-, and 6-year period. In other words, the tool accounts for disparities in a variety of socioeconomic factors that would impact outcomes and promotes health literacy as a means of improving health outcomes.

Study Design

This randomized controlled study (ID: NCT054511939, Registered 04/15/2023 at clinicaltrials.gov) was approved by the institutional review board at our university (HIC# 2000032637) and used the CONSORT checklist when writing this report [29]. This study aimed to evaluate the effectiveness of the SDMT with a cohort representative of the general knee OA patient population. Therefore, all patients presenting to our university-affiliated adult reconstruction orthopaedic surgery clinics with OA-associated knee pain consistent with OA diagnosis were included in this study. Exclusion criteria were previous history of inflammatory disease diagnosis, signs of cognitive impairment, and lack of English reading proficiency. Additionally, patients with a history of any type of arthroplasty, not limited to knee and including contralateral knee arthroplasty, were excluded. Age inclusion criteria was from 45–65 years of age to all ages. This was done to include a study population that was representative of patients seeking care. One patient was excluded from the intervention arm as none of the treatments offered to her by the clinician were listed within the tool. Additionally, patients who received consultation for knee arthroplasty revision were excluded. The control group received the standard of care visit from the clinician (normal office visit without use of the tool) while the intervention group received the same visit with the addition of the patient specific SDMT. Patients were randomized to the control arm (N = 55) or the intervention arm (N = 49) using a predetermined and algorithmically constructed assignment order.

Interventional Clinical Visit Pedagogy

Clinic visits for this study were conducted by a team of three orthopaedic surgeons and three advanced practice providers (APPs) specializing in OA management. Additionally, these clinicians were trained by study coordinators on how to customize and deploy the SDMT. To initiate the clinic appointment, study coordinators administered 3 PROMIS surveys (v2.0 Mobility, v1.1 Pain Interference, v1.2 Physical Function) to patients in both study groups to evaluate mobility, pain interference and physical function [30]. Additionally, patients in both study groups received a 5-question pre-survey to assess their understanding of their disease progression and their attitudes toward various treatment options (Table 2). Before the visit, all study-eligible patients (both cohorts) were shown an educational video clip that explained the etiology of knee osteoarthritis, nonoperative and surgical treatment options, and an animation of total knee arthroplasty surgery. For those in the intervention arm, during the office visit, clinicians selected two treatment options in the SDMT that best aligned with their clinical recommendations based on the patient’s knee OA progression. Non-operative conservative treatment options included exercise, hyaluronic acid injections, steroid injections and unloader braces. Knee arthroplasty was offered as a surgical treatment. The tool would then provide patients with predictions of future pain development, future level of activity, and future lost income over a five-year period dependent on the type of intervention recommended and ultimately pursued (Fig. 1). Clinicians then discussed the SDMT output with the patients before coming to a shared decision regarding a plan of care. Immediately following the visit, patients in both cohorts were queried again, with a five question post-survey, regarding their understanding of their disease progression, attitudes toward different treatments, and the impact the visit had on their decision-making. Standard of care patients followed the same protocol as intervention patients, but without the use of the SDMT predictive tool and subsequent discussion during the visit. Standard of care patients took the same post-visit survey.

One-Month Survey

All participants received a survey via email one month following their encounter. Patients were surveyed regarding their understanding of their disease progression, their attitudes toward (1) non-operative management, (2) injection-based treatments, and (3) TKA, and how impactful their visit with the clinician was in helping them determine their treatment plan.

Participant Groups

Study recruitment remained active from April 5, 2023, through July 25, 2023. A total of 104 patients provided informed consent and were enrolled on the day of their clinic appointment by study coordinators prior to their regularly scheduled visit with the healthcare clinician. From the original cohort, 15 control patients and 20 SDMT patients responded to the one-month follow-up survey. Demographic information for each group is reported in Table 1, with no significant differences in demographics between cohorts (p > 0.05). There were 55 patients in the SOC cohort and 49 patients in the SDMT intervention cohort.

Data Analyses

Data analysis was conducted using STATA statistical software (Version 17.0/BE). Demographic information was analyzed for both the SOC and SDMT groups, with two tailed Student t-tests with equal variance (Welch’s T-tests) being used for continuous variables such as age and PROMIS scores. The Chi-square test was used to determine if statistically significant differences existed between the control and intervention arm for categorical variables. Fisher’s exact test was used in the case that the observation number was less than 5. Lastly, two-tailed binomial proportionality t-tests were used to investigate differences in patient attitudes pre and post the visit (Table 2; Table 3). Statistical significance was defined as a p-value less than 0.05. In Table 2, the same response category was compared between cohorts using two-tailed binomial proportionality tests to investigate differences between the cohorts on pre- and post-visit surveys. Proportions for the pre-visit survey were then compared to the corresponding response category on the post-visit survey within the same cohort to determine general changes in attitude of the two groups throughout the course of the clinic visit (Table 3).

Results

Patient Demographics

Differences in patient demographics such as age, sex, race, ethnicity, as well as differences in the 3 PROMIS scores measured were not statistically significant between the control and SDMT groups (Table 1, p > 0.05). PROMIS scores demonstrated that both standard of care and SDMT groups had impaired mobility, decreased physical function, and increased pain interference pre-visit and preoperatively (Table 1). These mean measurements can be interpreted as “moderate” in terms of severity in accordance with HealthMeasures set cutoff points for PROMIS surveys [30]. The scores also confirmed that the subject population experienced effects from their OA diagnosis that were distinctly different from the general population [30].

Pre- and Post-Visit Survey Comparison

There were no significant differences between the two groups pre-visit according to their respective pre-visit survey results (Table 2). Post-visit, the proportion of patients stating that they understood the progression of their disease “very well” increased by 32 and 20 percentage points in the control and SDMT groups, respectively, when compared to pre-visit proportions, both statistically significant increases (Table 3, p < 0.05). Among attitudes towards treatment, the proportion of patients in the SDMT arm who responded that they were “extremely likely” to pursue nonoperative, conservative management was 26 percentage points greater than the control arm (p < 0.05) post-visit (Table 3).

Post-Visit and One-Month Survey Comparison

Comparing the results of the post-visit survey to the one-month follow-up survey, significant increases (p < 0.05) were observed when comparing the proportion of SDMT respondents who understood their disease “generally well” and who were “extremely unlikely” to pursue arthroplasty (Table 4). A significant decrease was observed for both groups from the post-visit survey to the one-month survey for the proportion of patients that understood their disease “very well” (Table 3–4, p < 0.05). The proportion of patients who were “somewhat likely” to pursue corticosteroid or HA injections was significantly greater in the SDMT group when compared to the control group at one-month (Table 3–4, p < 0.05).There was a significant difference in the proportion of patients who were “somewhat likely” to pursue non-operative treatment between the two groups at one-month, with 47% of the control population compared to 15% of the SDMT respondents (Table 4, p < 0.05), where 70% of SDMT respondents highlighted they would be “extremely likely” to pursue non-operative treatment at one-month, compared to 53% for the control group.

Discussion