Perceptions of Shared Decision Making Among Patients with Spinal Cord Injuries/Disorders

The present study was part of a large-scale evaluation to assess PCC in Veterans Affairs (VA). A cross-sectional mailed survey was conducted with individuals with SCI/D who were at least 1 year post injury and had received care at a VA health care facility or SCI Model Systems facility in the Chicago area in the previous 6 months. Participants received an initial survey packet, which included a cover letter describing the purpose of the study, a measure of demographics and injury characteristics, the Veterans' RAND 12-item Health Survey (VR-12), the Combined Outcome Measure for Risk Communication and Treatment Decision-Making Effectiveness (COMRADE, a measure of SDM), and a stamped business reply envelope. A follow-up mailing was sent to nonrespondents approximately 4 weeks later. This study was reviewed and approved by the Northwestern University Institutional Review Board.

Demographics included gender, age, race, education, marital/relationship status, living arrangement, distance and travel time to health care facility, and Internet use. Participants completed the VR-12, which assesses physical and mental health status.32 Higher scores indicate better physical and mental health, respectively. To assess SDM, participants completed the COMRADE, a 20-item scale, with 10 items created to evaluate risk communication (referred to here as “communication”) and 10 items to evaluate decision-making effectiveness (“effectiveness”).33 Communication refers to the patient's satisfaction with the information exchange about risks and benefits of a particular treatment or health decision; for instance, one communication item states, “The doctor gave me the chance to ask for as much information as I needed about the different treatment choices available.” Effectiveness refers to the patient's satisfaction with the outcome of the decision-making process; for instance, one effectiveness item states, “The decision shows what is most important for me.” COMRADE is a composite instrument using preexisting SDM measures, and it was developed through a multiphase study, including patient focus groups and interviews throughout development and revision.33 The measure has been shown to have consistent factor structure across time periods33 and acceptable levels of internal consistency.34 Scores on both subscales range from 0 to 100; higher scores indicate better SDM.

A randomly assigned identification number linked surveys to participants; patient name and social security number were used to obtain VA or Model Systems administrative data for each respondent. These data included SCI/D characteristics (level of injury, completeness of injury, etiology, time since injury).

We computed descriptive statistics (frequencies, means and standard deviations) and bivariate analyses (t tests, chi-square, Pearson's correlation, as appropriate) to examine demographic, health, and SCI/D characteristics related to SDM. To explore factors independently associated with SDM, multiple linear regressions were performed for each of the COMRADE subscale scores; models were developed using the variables found to have significant bivariate relationships as predictors, after controlling for key demographic variables entered as covariates: gender, age, race (White vs other), group (Veteran vs civilian), marital status (married or member of couple vs other), and education (high school or less vs some college or more).

A total of 451 Veterans with SCI/D were mailed surveys; 21 were returned as undeliverable, and 5 Veterans were deceased, leaving a total sample size of 425 Veterans with SCI/D. Additionally, 825 non-Veterans (civilians) with SCI/D were mailed surveys; 8 were excluded as ineligible (eg, under 18 years of age), 18 were deceased, and 64 were returned as undeliverable, leaving a total of 735 civilians with SCI/D. The final sample of respondents (N = 450) included 124 Veterans with SCI/D (response rate = 29%) and 326 civilians (response rate = 44%) with SCI/D.

Respondents did not significantly differ from nonrespondents in terms of marital status, gender, etiology of the injury/disorder, level of injury/disorder, or completeness of injury/disorder. Respondents were significantly more likely to be White (74.9% vs 61.1%; p < .001), significantly less likely to be Black (22.2% vs 34.7%; p < .001), and significantly less likely to be of Hispanic/Latino ethnicity (4.9% vs 8.9%; p < .001). Respondents were also significantly older than nonrespondents (M = 53.2 years, SD = 14.48 vs M = 50.1 years, SD =16.47; p = .002), and tended to have had their injury longer (M = 19.3 years, SD = 12.37 vs M = 16.3 years, SD = 13.62; p = .0003). Overall, rates of missing data for study variables were low, ranging from 0.0% to 4.7%. However, effectiveness and communication total scores could only be computed for 90.7% and 90.2%, respectively, because total scores cannot be computed without complete data.

Participants were predominantly male (82.6%) and White (70.3%) and were, on average, 53.5 years of age (SD = 14.30). Veterans with SCI/D were significantly more likely to be male (94.3% vs 78.2%; p < .0001) and Black (31.2% vs 18.4%; p = .004) and less likely to be of Hispanic/Latino ethnicity (2.5% vs 7.7%; p = .043). Veterans also tended to be older (M = 60.6 years, SD = 12.09 vs M = 50.8 years, SD = 14.16; p < .0001) and lived farther from their health care facility (M = 45.4 miles, SD = 61.66 vs M = 14.5 miles, SD = 397.24; p < .0001). Veterans were more likely to have been to a doctor or hospitalized in the 6 months prior to the survey (87.9% vs 78.8%; p = .0274). In terms of SCI/D characteristics overall, 43.6% had paraplegia and had their SCI/D on average for 19.2 years (SD = 12.37). Veterans with SCI/D were significantly more likely to have an incomplete injury (75.0% vs 45.0%; p < .0001), and to have a nontraumatic etiology (eg, cauda equina syndrome, nonmalignant neoplasms), due to the fact that all civilians sampled had a traumatic etiology (37.0% vs 0.0%; p < .0001) (see Table 1).

Overall effectiveness (M = 63.9, SD = 15.50) and communication (M = 57.7, SD = 17.06) scores were close to the midpoint. Bivariate analyses were conducted using the 2 COMRADE subscales as outcomes. No differences were found between Veterans and civilians for either of the COMRADE subscale scores, and none of the demographic variables (eg, gender, race, education, age) were related to effectiveness or communication scores. Level of injury was significantly related to effectiveness scores; individuals with tetraplegia had higher scores (M = 65.1, SD = 15.80) than individuals with paraplegia (M = 61.9, SD = 15.14; p = .04) (see Table 2). Additionally, effectiveness was positively, though weakly, correlated with physical health, (r = 0.12, p = .002) and moderately correlated with mental health (r = 0.27, p < .001). Communication score was also positively and moderately correlated with mental health (r = 0.24, p < .001) (see Table 3).

Two multiple linear regressions were performed. The first regression used effectiveness as the outcome and level of injury, physical health, and mental health as predictors; even after controlling for covariates, all of these variables had significant independent effects on effectiveness, F(9, 377) = 4.81, p < .001, R² = 0.103. [Given that these covariates were not significant, a separate multiple linear regression was performed using only the predictor variables; results were similar, F(3, 390) = 13.68, p < .001, R² = 0.095.] The inclusion of level of injury, physical health, and mental health resulted in a significant increase in explained variance over the model only including covariates, F(3, 377) = 13.33, p < .001, ΔR² = 0.095. Individuals with tetraplegia, better physical health, and better mental health had higher effectiveness scores (see Table 4). The second regression used communication as the outcome and mental health as the predictor, as this was the only variable significantly impacting communication scores; even after controlling for covariates, individuals with better mental health status had higher communication scores, F(7, 385) = 4.00, p = .003, R² = 0.068. [As above, a separate linear regression was conducted without covariates; similar results were obtained, F(1, 400) = 24.84, p < .001, R² = 0.059.] The inclusion of mental health resulted in a significant increase in explained variance over the model only including covariates, F(1, 385) = 24.93, p < .001, ΔR² = 0.060 (see Table 5).

New directions in health care delivery encourage greater use of SDM.35,36 Further, research suggests that SDM does not adversely affect patient outcomes34 and may in fact improve outcomes such as adherence to treatment21 and satisfaction with care.20 Though past research14,24,25 has shown an impact of personal characteristics, such as age and race, on SDM, these relationships were not significant in the present study. However, the effect sizes obtained in past research were generally small, except for marriage, which had a moderately strong relationship.14 Instead, our study found that select SCI/D and health characteristics were the only variables associated with SDM. Specifically, individuals with tetraplegia, better physical health, and better mental health were shown to have higher perceptions of decision-making effectiveness, and individuals with better mental health were shown to have higher perceptions of risk communication.

Individuals with tetraplegia demonstrate higher health care utilization and presence of other health conditions.2–5 They may require greater autonomy, due to the amount of self-care in which they engage, which may explain why they demonstrated higher effectiveness scores in the present study. Though Engkasan and colleagues24 demonstrated that individuals with SCI/D preferred being actively involved in coping with their health conditions and making treatment decisions, they did not compare individuals with paraplegia and individuals with tetraplegia. However, providers in their study felt that, in certain situations (eg, the patient is very elderly or severely disabled), another party, such as the provider or family caregiver, should make treatment decisions instead. The informal caregiver role was viewed as especially important in situations where the caregiver would be responsible for assisting the patient, such as in bladder or bowel care programs.24 This finding suggests that individuals who are more dependent on caregivers, such as individuals with tetraplegia, may be less involved in SDM. Therefore, it is unclear why, in our study, we found greater perceptions of decision-making effectiveness among individuals with tetraplegia. More research is needed to determine why effectiveness scores were higher in this group and how to use that information to increase SDM. Whereas individuals with paraplegia were as satisfied with the information exchange as individuals with tetraplegia, they were less satisfied with the outcome of the decision-making process. Individuals with paraplegia may benefit from efforts to encourage a stronger decision-making process. Health care facilities may also consider including SDM training to patients and providers.

Health status was also associated with SDM. Poorer physical health was associated with lower effectiveness scores, and poorer mental health was associated with low effectiveness and communication scores. Individuals with more physical and/or mental health conditions may have greater health care utilization, resulting in more potential decisions about treatment.37 The current findings suggest that patients with mental and/or physical health issues are not engaging in an effective SDM process with their providers and that individuals with mental health issues also lack sufficient communication about risks/benefits of treatment. It is unclear whether patients do not feel empowered to engage in SDM or whether providers are following the more traditional, paternalistic format rather than one involving SDM.6 Either of these may occur for a variety of reasons, such lack of time for the provider to educate the patient or because the patient prefers the provider to make the decision. Additionally, as individuals with complex health care needs are more likely to experience competing outcomes17 of different treatment options, such as a treatment option that alleviates symptoms of one condition but exacerbates symptoms of a comorbid condition, patient values and preferences become even more essential in treatment decisions.18 

Providers may view individuals with poor health as less able to make treatment decisions, especially individuals with mental health issues.38 This may explain why, in this study, individuals with poorer mental health were found to have lower scores on both COMRADE subscales. However, a study among individuals with schizophrenia found positive effects of a SDM intervention in which nurses trained patients on treatment options and “planning talk” with their physicians.16 Additionally, contemporary approaches to mental health encourage greater autonomy and choice among individuals receiving mental health care.38 

One way to encourage patient autonomy is through decision aids, which are educational materials that compare treatment options and help guide patient decision making and fulfill the “risk communication” portion of SDM.39 However, it is often difficult to construct decision aids for patients with multiple conditions.17 Individuals with SCI/D experience many secondary conditions and complications that may make it especially difficult to create a suitable decision aid. As such, it is important for providers to actively engage in SDM with patients during health care encounters through communication about risks and benefits of treatment options and cooperative decision making. Col and colleagues40 offer a framework containing 5 skills providers should possess to effectively engage in SDM: understanding the concept, communication skills, understanding interprofessional sensitivities, understanding the roles of different providers, and additional relevant skills, including identifying available resources and understanding and modifying barriers.

Though individuals with SCI/D may be uninterested in SDM, past research suggests this is not the case.24,30,31 Low communication and/or effectiveness scores could instead be due to provider behavior. Providers may lack the time to engage in SDM in care visits. A systematic review of barriers to SDM found that time was the most frequently cited concern by providers.41 However, studies examining length of consultations with or without SDM have generally found no differences.42,43 Patients and providers may also lack training on engaging in SDM. Many recent studies on SDM have evaluated provider interventions. Edwards and colleagues34 evaluated an intervention to improve SDM among general practitioners. Though the intervention did not result in significant changes in patient outcomes, they did find improvements in patient perceived risk communication and patient confidence. Further, a systematic review of randomized controlled trials on SDM interventions showed some improvements in patient knowledge and satisfaction; 2 studies that included individuals with mental health issues showed positive effects of the interventions.39 Additionally, SDM is most effective when it builds on a strong, preexisting relationship between patient and provider.10,44 This highlights the importance for SCI providers to build and maintain strong rapport with their patients.

Because this study used a cross-sectional survey, response and recall biases are possible. Every effort was made to maximize response rates, including reminders and replacement surveys to nonrespondents. Response rates were slightly higher for civilians than Veterans. This is likely because the Model Systems facility was able to provide a larger incentive to its participants. Further, though we found significant relationships between SDM and SCI/D and health characteristics, there may be other significant variables that were not measured, such as household income. However, this study provides important results that can be explored in future research, such as examining why individuals with tetraplegia demonstrated better effectiveness scores. Finally, we selected the COMRADE because it is a composite instrument assessing SDM constructs that were identified as most important by patients; this measure has not been separately validated among individuals with SCI/D. Very little research has been conducted on SDM among individuals with SCI/D, and studies have been mainly qualitative. Though COMRADE is worded generally, to apply to a variety of clinical populations, additional research is needed to determine whether another measure of SDM would more effectively measure this concept among individuals with SCI/D.

Though past research found demographic characteristics, such as age and race, impact perceptions of SDM, in the present sample of individuals with SCI/D, SCI/D and health characteristics had stronger associations with SDM. More research is needed to understand SDM in these populations, particularly how it occurs in specific patient-provider interactions. Interventions to increase engagement in SDM, especially for individuals with paraplegia, poorer physical health, and poorer mental health, may also be beneficial. This represents an opportunity for providers and health care organizations to target improving SDM in these subgroups. SCI providers may also benefit from SCI/D-specific findings in the present study and improve incorporation of SDM into care.

The study was conducted in accordance with ethical guidelines and was reviewed and approved by the Northwestern University Institutional Review Board. This paper reflects only the authors' opinions and does not necessarily reflect the official position of the Department of Veterans Affairs or the United States Government. The authors declare no conflicts of interest related to this manuscript. Study funding was provided by the Department of Veterans Affairs, Office of Patient-Centered Care and Cultural Transformation and Health Services Research & Development Quality Enhancement Research Initiative (PEC 13-002; PI: Sherri LaVela, PhD, MPH, MBA).