Association of Preferences for Participation in Decision-making With Care Satisfaction Among Hospitalized Patients

This survey study investigates the association of hospitalized patients’ desire to delegate decisions to their physician with care dissatisfaction.


Introduction
Greater patient participation in medical decisions improves satisfaction 1 and objective health outcomes 2 while fostering clinical care more strongly aligned with patient preferences. 3 Accordingly, numerous initiatives encourage patients to actively participate in decisions, including the Affordable Care Act. 4 Patient preference for engagement in medical decision-making varies across individuals, subgroups, and decision types. Patients generally want to receive information 5 but often wish to delegate decisions to physicians, [6][7][8][9] a preference that may be stronger among older patients. 8,10 Patient preferences for engagement in decisions may differ based on technical components of the decision and subjective aspects of the outcomes. 1,5 Variation in patient preferences and expectations concerning engagement may be associated with patient-reported quality metrics, including satisfaction, 11 ratings of care, 12 quality of life, 13 and other measures of health service quality. 14 Because preferences and expectations exhibit geographic variation, 15 such associations could complicate the interpretation of patient-reported outcomes (PROs) as measures of clinician performance. Despite the growing use of such performance metrics, 16 literature on the association of PROs with preferences for participation in decision-making is limited, particularly among urban minority racial/ethnic populations. Because race/ethnicity has important associations with patient trust 17 and satisfaction, 18 studies that include diverse populations are essential to understand preferences among diverse groups and their implications for PROs.
Prior research regarding the consequences of patients' desire to participate in health care decisions has been limited by small sample sizes, lack of adjustment for important confounding factors, and selection bias resulting from restriction to educated populations with sufficient ability to answer questionnaires. 19,20 We studied the association of preferences for participation in clinical decisions with care satisfaction among a predominantly minority racial/ethnic sample of hospitalized patients at The University of Chicago Medical Center, an urban academic medical center.

Data Sources and Study Population
This survey study in an academic research setting used data collected through the University of Chicago Hospitalist Study, 21  guideline. Trained research assistants obtained written informed consent from all admitted patients immediately after admission for an inpatient survey with 44 questions, including race/ethnicity, educational attainment, general self-assessed health status, and patient preference for medical decision-making. We contacted patients 30 days after discharge to ask 56 follow-up questions, including postdischarge medical care use, health status, and service quality measures during the previous hospitalization.

Data Elements
The following 3 service quality measures were obtained from the 30-day follow-up survey: overall rating of care received (excellent, very good, good, fair, or poor), satisfaction with physician care (extremely satisfied, somewhat satisfied, somewhat dissatisfied, or extremely dissatisfied), and confidence and trust in the physicians providing treatment (yes always, yes sometimes, or no). Poor service quality measures were defined as follows: dissatisfaction with overall care (fair or poor), physician care (extremely or somewhat dissatisfied), and no confidence and trust (vs yes always or sometimes).
The primary independent variable was patient-reported preference to leave medical decisions to their physician derived from a survey item ("I prefer to leave decisions about my medical care up to my doctor") to which patients responded as definitely agree or somewhat agree vs somewhat disagree or definitely disagree. Additional detail regarding this variable has been previously published. 8,9 Other patient-level independent variables include the following: age, sex, race/ ethnicity, educational attainment (any high school or less, high school graduate, some college or junior college, college graduate, or any graduate-level educational attainment), insurance type (private, Medicare, Medicaid, or no insurance), general self-assessed health status (excellent, very good, good, fair, or poor), Charlson Comorbidity Index score, and transfer from another health care facility or service. The main outcomes were patient-reported dissatisfaction with overall service, dissatisfaction with physician care, and lack of confidence and trust in the physicians providing treatment, which were obtained from the 30-day follow-up survey.

Statistical Analysis
For the bivariate analyses, we used t test (age) and Pearson χ 2 test to examine the differences in patient characteristics and service quality measures with statistical significance prespecified at P < .05. To investigate the association of patient preference for medical decision-making with the 3 service quality measures, we specified multivariable logistic regression models, with the 4-level preference indicator (definitely agree or somewhat agree vs somewhat disagree or definitely disagree) as the primary independent variable (with definitely agree as the reference category). The independent covariates were the patient-level variables described previously and categorical variables for year, month, attending physician, the 10 most frequent principal diagnoses, and admission on a weekend. The dependent variables of each regression were reverse-coded derived from the following service quality measures: dissatisfaction with overall service (rating of fair or poor vs excellent, very good, or good), dissatisfaction with physician care (somewhat dissatisfied or extremely dissatisfied), and lack of confidence and trust in the physicians providing treatment (vs yes always or yes sometimes).
A small amount of missing data (5.8% for educational attainment, 2.8% for insurance type, 5.0% for general self-assessed health status, 8.2% for patient preference for medical decisionmaking, 4.1% for overall rating of care received, 3.2% for satisfaction with physician care, and 3.7% for confidence and trust in the physicians providing treatment) was imputed using multiple imputation methods with 20 iterations, which yielded consistent estimates with valid inference in estimation. 22 We used an ordered logit model for ordinal variables (educational attainment, general self-assessed health status, patient preference for medical decision-making, overall rating of care received, satisfaction with physician care, and confidence and trust in the physicians providing treatment) and a multinomial logit model for the nonordinal variable (insurance type). To confirm the robustness of the imputations, we compared the results with and without imputations. Statistical analyses were performed with Stata, version 11.2 (StataCorp LLC).

JAMA Network Open | Medical Education
Participation in Decision-making and Care Satisfaction in Hospitalized Patients

Results
The sample population included 13 902 admissions for which both the initial inpatient and 30-day follow-up surveys were completed. Between July 1, 2004 completed the inpatient survey. The others were discharged without completing the survey.
Compared with participants who responded to the inpatient survey, nonparticipants were older, were more likely to be African American, were less healthy, and had a shorter length of stay and reduced total hospitalization cost. Among the inpatient survey participants, 13 902 of 22 954 patients (60.6%) completed the 30-day follow-up survey. For the 30-day follow-up survey, nonparticipants were more likely than participants to be male, less educated, and less healthy and were more likely to be publicly insured or uninsured. Table 1 lists patient sociodemographic characteristics, general self-assessed health status, and patient preference for medical decision-making. The mean (SD) patient age was 56.7 (19.1) years, 60.4% (n = 8397) were women, 39.6% (n = 5505) were men, and 74.2% were African American.
Overall, 53.2% had no higher educational attainment, 22.7% were insured by Medicaid, and 51.1% reported a general self-assessed health status of fair or poor. The proportions of respondents who agreed and disagreed with delegating decisions to the responsible physician were 71.1% and 28.9%, respectively. A statistically significantly higher proportion of those who agreed rated their overall care as excellent or very good compared with those who disagreed (68.0% vs 62.5%; P < .001).
Similarly, a statistically significantly higher proportion of those who agreed were extremely satisfied with the physician care received (67.8% vs 62.5%; P < .001). Table 1 also lists the responses to the 3 service quality measures: 9.6% evaluated the overall rating of care received as fair or poor, 6.5% expressed dissatisfaction with physician care, and 4.9% did not have confidence and trust in the physicians providing treatment. Figure 1 shows that the characteristics of age younger than 65 years, White race, higher educational attainment, and worse general self-assessed health status were associated with greater degrees of dissatisfaction in the overall rating of care received. White race, higher educational attainment, and worse general selfassessed health status were also associated with greater degrees of dissatisfaction with physician care. The characteristics of age younger than 65 years and worse general self-assessed health status were associated with greater lack of confidence and trust in the physicians providing treatment. Figure 2 shows the association of patient preference for medical decision-making with patientreported measures of dissatisfaction and lack of confidence and trust in the physicians providing treatment. Reduced willingness to defer medical decisions to the physicians providing treatment was associated with greater dissatisfaction in the overall care received, more dissatisfaction with physician care, and reduced confidence and trust in the physicians providing treatment. Compared with patients who definitely preferred to leave medical decisions to their physician, those who definitely disagreed with delegating decisions to their physician were more likely to be dissatisfied in the overall care received (14.2% vs 7.3%), express dissatisfaction with physician care (9.2% vs 4.9%), and lack confidence and trust in the physicians providing treatment (7.9% vs 3.9%) (P < .001 for all).

JAMA Network Open | Medical Education
Participation in Decision-making and Care Satisfaction in Hospitalized Patients

Discussion
The core elements of shared decision-making are physician sharing of information and patient participation in decisions, which may improve patient satisfaction and health outcomes. 23 However, in the present study, a desire to participate in decisions was associated with reduced satisfaction and less confidence and trust in the physicians providing treatment. Prior literature has found that patient preference for involvement in decisions is heterogeneous 7,19 and is inconsistently associated with outcomes 24 and that physicians may overestimate patients' desire to make decisions. 25 Given this finding and the increasing use of PROs in performance reporting, understanding the implications of patients having a larger role in decisions for PROs is critical. Although we cannot know with certainty the reasons for the observed association between a greater desire to participate in decisions and reduced satisfaction and trust, we hypothesize that an increased desire to participate in decision-making is associated with greater expectations of care and communication. 26     Reduced willingness to defer medical decisions to the treating physician was associated with greater dissatisfaction in overall rating of care received, greater dissatisfaction with physician care, and reduced confidence and trust in the treating physicians.
The medical community's desire for greater patient participation in decisions arises from a theory based on sociopolitical factors, 24 including cultural predispositions for patient autonomy 6,28 and changing views regarding the ethical aspects of decision-making. 1 However, there is some evidence for greater satisfaction and improved outcomes as a result of patient participation in decisions. 19,20 Although definitive research in this area is limited, the discrepancy between prior literature and the present study may be worth exploring. First, much literature has studied healthy volunteers or outpatients. 3 Based on a sample of hospitalized patients, our findings may reflect a vulnerability that modifies the consequences of preferences for participation on care satisfaction and trust.
Second, many previous studies on this subject have been based on middle-class or upper-class predominantly White populations, 3,18,32 in contrast to this urban African American sample.
Preferences for care and decision-making differ markedly across patient populations and over time. 7 These considerations may explain in part why we did not observe greater distrust among the African American patients in this study, as other studies have. 29 Furthermore, distrust is associated with receipt of care inconsistent with strongly held patient preferences and with lack of patient centeredness. 29  to a desire to participate in decisions and create more opportunity for dissatisfaction. 12 We cannot know with certainty the mechanism of the associations found in this study; however, based on theory and previous literature, 32,33 we believe that higher expectations of care and communication that arise among assertive patients with a greater desire to participate in decision-making mediate the findings that we observed.

Implications for Future Research
There are several important implications of this study. The associations we found suggest that efforts to involve patients in shared decision-making should be sensitive to the heterogeneity of patients' preferences to participate and encourage only types of involvement concordant with their wishes.
This implication supports concerns in the published literature suggesting caution regarding overzealous efforts to involve patients because many do not wish to have active participation. 7 Efforts to involve patients in health care decisions should be individualized and reflect the importance of shared decision-making in maximizing satisfaction for a substantial proportion of patients. However, our subgroup analyses in Figure 1  In addition, performance reporting may create a disincentive for hospitals to encourage an environment of shared decision-making. In this study, patients with fair or poor general self-assessed health status were consistently more dissatisfied than those with better health across the 3 service quality measures. In addition to the consequences that this association may have for PROs at institutions with a disproportionately large patient base with poor health, critical access hospitals and other hospitals with low operating margins tend to take care of sicker patients. 36 Because poor ratings of satisfaction are associated with higher rates of emergency department use, 37 patients in poor health who are also dissatisfied with care received may be large contributors to statistics on ambulatory-sensitive conditions.

Limitations
This study has limitations. First, the results of this study at a single medical center may not be generalizable outside of the urban, largely African American population we studied. Second, 29.3% of admitted patients did not participate in the survey. However, although there were certain differences in baseline characteristics between participants and nonparticipants, the magnitude of these variations was small. Therefore, selection bias is unlikely to have impacted our primary findings.
Third, we studied hospitalized patients, for whom acute illness may modify satisfaction and confidence and trust. This is in contrast to the absence of acute illness in the outpatient setting, where physician-patient relationships are more durable and there is greater continuity of care. In addition, dissatisfaction may be associated with a lack of willingness to defer decisions. However, because patients were asked about their preference for participation soon after hospitalization, the temporal association suggests that these preferences would be based on prior health care experiences and innate patient characteristics, rather than dissatisfaction with care received after survey administration. Also, prior experience may alter patient satisfaction with current health care.

JAMA Network Open | Medical Education
Participation in Decision-making and Care Satisfaction in Hospitalized Patients However, satisfaction during a current or recent hospitalization is the strongest determinant of patient satisfaction with care. 32,33 Fourth, although shared decision-making requires participation of both physicians and patients, we do not have information on physician preferences to engage in shared decision-making. However, because physicians on the general internal medicine service at The University of Chicago Medical Center are assigned to patients based on a fixed call schedule, the decision-making preferences of physicians should be similar across categories of patient preference for medical decision-making. The inclusion of attending physician in our multivariable logistic regression models also minimizes potential consequences of physician preferences, although we cannot know with certainty how physician behavior might differ according to physician-patient relationships.
Fifth, we have not measured actual participation (but rather preferences for such) or other contributing factors (eg, personality) and cannot draw definitive conclusions regarding the mechanism of the observed associations. Although we used delegation of decisions as a proxy measure of desire for shared decision-making, future research may evaluate how robust the observed associations are to alternative measures of shared decision-making.
Sixth, despite potentially important associations, 38 we did not measure the consequences of participation in decision-making on objective health outcomes. If a preference for participation in medical decisions improves health outcomes, such benefits may outweigh the satisfaction-reducing associations we observed with PROs. Patient participation in decision-making may help physicians exercise responsible stewardship, which may reduce unnecessary resource use while maintaining or improving outcomes.

Conclusions
This survey study provides insights regarding the heterogeneous association that patient