Racial and Ethnic Differences in Shared Decision Making Among Patients With Hypertension: Results From the RICH LIFE Project

Abstract

Background

Racial and ethnic disparities in hypertension care persist. Shared decision making (SDM) is promoted in hypertension guidelines. However, evidence is lacking on how race, ethnicity, and SDM relate, and the effect of SDM on hypertension control in diverse groups. We aimed to explore the relationships among SDM, blood pressure (BP), race and ethnicity, and other decision‐making factors in patients with hypertension.

Methods and Results

Longitudinal analysis of data from the RICH LIFE (Reducing Inequities in Care of Hypertension: Lifestyle Improvement for Everyone) project's participants (n=1426) with uncontrolled hypertension was performed using descriptive statistics, linear regression, and generalized estimating equations. Participants were middle‐aged (mean age 60±11.6 years), predominantly women (59.4%, 847), non‐Latino Black (59%, 844), and high school graduates or below (65%, 931). The mean SDM score was 7.2±2.6 out of 9, and the mean baseline systolic and diastolic BP were 152.2±12.0 and 85.3±12.1 mm Hg. Non‐Latino Black people had 0.14 points higher mean SDM score (P<0.001) than non‐Latino White people. Systolic BP reduction over 12 months was greater with a higher SDM mean score (β=−0.42, P=0.035). Baseline characteristics associated with SDM included more than high school education (β=−0.08, P=0.045), hypertension knowledge (β=−0.05, P=0.046), considering taking BP medication as very important (β=0.06, P=0.022), and patient activation (β=0.09, P=0.001).

Conclusions

There was greater BP reduction for patients with higher SDM score at follow‐up, and associations between SDM and race and ethnicity, education, hypertension knowledge and attitude, and patient activation. Future research should further explore SDM differences among racial and ethnic groups to better align hypertension care with patients' needs.

Nonstandard Abbreviations and Acronyms

RICH LIFE

Reducing Inequities in Care of Hypertension: Lifestyle Improvement for Everyone

SDM

shared decision making

Clinical Perspective.

What Is New?

• This article reports that systolic blood pressure reduction over 12 months was greater with a higher shared decision making (SDM) mean score.

• In the context of hypertension care, Black people, individuals who consider taking blood pressure medication as very important, and those with higher patient activation were more likely to have high SDM.

• Participants with higher education and hypertension knowledge were less likely to have high SDM.

What Are the Clinical Implications?

• Our study findings emphasize the importance of SDM in clinical practice to achieve better hypertension outcomes.

• Factors such as education level, hypertension knowledge, patient's outlook on medication use, and patient activation can be crucial for tailoring hypertension interventions to specific patient populations.

• Our results also show an association between SDM and patients' perceived importance of taking medication, further highlighting the opportunity to use SDM strategies to promote medication adherence to improve hypertension control.

Hypertension is a major risk factor for heart disease and stroke, contributing to about 1800 deaths per day ¹ and at estimated annual costs of $51 billion per year in the United States. Although hypertension control rates have been improving over the previous decades, recent national control rates have declined. ² Racial and ethnic disparities persist, with hypertension prevalence higher among Black (57%) and Latino (45%) than White people. ³ Adherence to antihypertensive medication, an important determinant of hypertension control, is also lower among Black (36%) and Latino (34%) compared with White people (24%). ⁴ To date, these disparities have been attributed to multilevel factors that influence hypertension care and outcomes, including issues on access to medications, and inadequate hypertension awareness and treatment. ⁵ , ⁶

Current models of care are failing to meet the needs of many patients with hypertension. Patient care continues to evolve from paternalistic models to patient‐centered approaches. ⁷ , ⁸ Shared decision making (SDM) is a patient‐centered strategy recommended in national and cardiovascular primary prevention guidelines to improve hypertension control. ⁹ , ¹⁰ , ¹¹ SDM happens when clinicians and patients make decisions together using the best available evidence along with patients' preferences and values. ¹² In the context of hypertension, SDM grants explicit mention of the decision to be considered (what type of hypertension treatment), the options (pharmacological and nonpharmacological/lifestyle changes), and the risks and benefits of the options. Consequently, SDM can be particularly important in hypertension management due to the various treatment options, each with their own risk/benefit profile, including antihypertensive medications being generally lifelong, ¹³

Interventions to increase SDM have been associated with increased use of preventive care, fewer or shorter hospitalizations, decreased use of invasive treatments, lower medical costs, as well as improvements in clinician–patient interaction, patient satisfaction, health literacy, and adherence to treatment. ¹⁴ , ¹⁵ Patients also have better longer‐term health outcomes across several illnesses, such as improvement in hemoglobin A1c, high‐density lipoprotein cholesterol, low‐density lipoprotein cholesterol, diabetes disease self‐efficacy and control, ¹⁶ , ¹⁷ and self‐care behaviors. ¹⁸ Evidence is also encouraging in hypertension. Subgroup analysis of a prospective controlled clinical study revealed that patients with high interest in SDM had a decrease in systolic (r=−0.49, P=0.042) and diastolic blood pressure (BP) (r=−0.524, P=0.043). ¹⁹ Another study also found that hypertension control was associated with a higher preference for SDM. ²⁰

Moreover, SDM is an important strategy to advance health equity. ²¹ Despite promising findings, the current evidence base is limited by sparse publications on SDM in the context of hypertension and noted underrepresentation of diverse populations ²² , ²³ in addition to a lack of outcomes reported by race and ethnicity ²⁰ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ and high risk of bias. ³¹ Efforts for the inclusion of SDM as part of standard care are increasing. ⁸ , ³² Nevertheless, there is little evidence to inform SDM specifically in hypertension, ¹⁵ , ³¹ , ³³ especially on racial and ethnic differences or similarities.

We adapted the Ottawa Decision Support Framework ³⁴ , ³⁵ to guide choice of relevant patients' clinical and nonclinical determinants related to health decisions. Our adapted model proposes that the determinants of decision influenced the degree to which SDM occurs, which in turn may impact health outcomes. Determinants of decision explored at baseline included sociodemographic (age, sex, race and ethnicity, education, income, employment, and marital status) and clinical characteristics (health status), hypertension knowledge and beliefs (hypertension knowledge, attitudes, and perceptions), and resources to make and implement decision (patient activation, health literacy). This framework ³⁴ , ³⁵ highlights how the selection of 1 option over another depends on other aspects that influence how the decision is perceived as well as personal resources available to make and implement the decision. An image of our adapted model was included in our previously published article. ³⁶

We also used the National Institute on Minority Health and Health Disparities research framework ³⁷ to inform our approach and to explore the relationships between minority health and other variables. The National Institute on Minority Health and Health Disparities framework supports the complexity of health disparities research and the conceptualization of the domains that influence health disparities. These domains are relevant for the promotion of minority health and comprise biological, behavioral, physical/built environment, sociocultural environment, and health care system. Discrimination at the individual, interpersonal, community, and societal levels might be present in the sociocultural environment and have the potential to affect cardiovascular outcomes. ³⁸ Given that patients from different racial, ethnic, and cultural backgrounds appraise their decision‐making process differently, ³⁹ the National Institute on Minority Health and Health Disparities framework was applied to examine possible health disparities influencing factors related to SDM in the context of hypertension. Influencing factors examined in this study include sociodemographic characteristics (age, sex, education, income, and employment), hypertension knowledge and beliefs, and health literacy. The National Institute on Minority Health and Health Disparities framework aided the exploration of disparities that can emerge among groups from different racial and ethnic backgrounds.

Given the nascent state of the knowledge in the area, the purpose of this study was to examine (1) baseline determinants of SDM and (2) relationships among race and ethnicity, SDM, and BP among patients with hypertension.

Our study used data from the RICH LIFE (Reducing Inequities in Care of Hypertension: Lifestyle Improvement for Everyone) project. ⁴⁰ RICH LIFE was designed to compare the effectiveness of enhanced standard of care with a clinic‐based multilevel intervention for improving BP control and patient activation, and reducing disparities in BP control. The results showed that adding a collaborative care team to enhanced standard of care improved patient ratings of chronic illness care. ⁴¹

METHODS

The data that support the findings of this study are available upon reasonable request from qualified researchers trained in human subject confidentiality protocols and may be sent to the Johns Hopkins Center for Health Equity at healthequity@jhmi.edu.

Design

Our team used a longitudinal analysis of data from the RICH LIFE project. ⁴⁰ RICH LIFE, a 2‐arm, cluster‐randomized pragmatic trial, used a pragmatic and sustainable approach combining multilevel strategies to reduce cardiovascular disease‐related risk factors. The intervention involved patient‐centered care and behavior change counseling provided by a nurse care manager. Additionally, when needed, participants also had the option to access the services of a community health worker, a group of specialists, or both, to improve BP control. The RICH LIFE project established partnerships with 30 clinical sites, including federally qualified health centers and private clinics across Maryland and Pennsylvania. To be included in the RICH LIFE project, patients had to have: (1) a diagnosis of hypertension with at least 1 other cardiovascular risk factor (eg, coronary heart disease, diabetes, hyperlipidemia, current tobacco smoker, or diagnosis of depression); (2) a BP reading ≥140/90 mm Hg at the last provider visit; (3) age ≥21 years; and (4) self‐identified as non‐Latino White, non‐Latino Black, or Latino. We hypothesized that, after controlling for intervention group assignment and covariates, (1) SDM scores would differ by patients' race and ethnicity, and (2) higher SDM scores would be associated with greater reduction in systolic BP over 12 months.

Procedures and Sample

The Johns Hopkins Medicine Institutional Review Board approved this study and all RICH LIFE project procedures (IRB00085630). Patient eligibility was determined based on electronic medical record data. Eligible individuals received mailed invitations to participate, along with an oral consent paper copy. Individuals who were interested were further screened, consented, and completed the baseline survey via phone. Participants were interviewed in English or Spanish, according to their native language and preference, by trained bilingual research assistants to confirm the accurate use and interpretation the language. Baseline survey data collection was performed between August 1, 2017 and October 31, 2019. The 12‐month follow‐up survey data were collected between August 7, 2018 and March 31, 2021.

Measures

Table 1 presents selected RICH LIFE measures included in the current study. Critical terminologies used throughout this article include the terms Latino and non‐Latino. Hispanic refers to Spanish‐speaking people who reside in the United States, especially individuals of Latin American descent. ⁴² However, there are many other languages and dialects spoken in Latin American countries. For example, the language spoken in Brazil is Portuguese, in Guyana and Belize is English, and Haiti and French Guiana is French, to mention a few. ⁴³ Consequently, the term Latino, which means individuals, men and women, of Latin American origin or descent, ⁴⁴ more inclusively represents the population examined in this article. Because gender diversity is not discussed in this study, the term Latinx was not used. ⁴⁵ , ⁴⁶ In addition, we used Centers for Disease Control and Prevention guidance for the preferred terms for select population groups and communities. ⁴⁶

Table 1.

RICH LIFE Measures Included in this Secondary Analysis

Measure (data source)	No. of items	Score range	Validity and reliability	Time point
Patient‐level clinical data (EMR)
Systolic BP	…	…	…	Baseline
Diastolic BP	…	…	…	Baseline
Patient‐reported data (telephone interview)
Health status: PROMIS global scale	10	9–55	Ordinal α=0.875 86	Baseline
Patient Activation Measure	13	13–52	Cronbach α=0.81 87	Baseline
Hypertension knowledge	1	0–1	Not reported	Baseline
Hypertension attitudes	1	0–1	Not reported	Baseline
Hypertension perceptions	1	0–1	Not reported	Baseline
Screening Questions for Limited Health Literacy	3	0–12	AUROC curve 0.72–0.80 53	Baseline
Shared decision making: CollaboRATE	3	0–9	r=0.79 and ICC=0.86*, 88	12 mo
Sociodemographic measures (telephone interview)
Age, sex, race and ethnicity, education, income, employment, marital status	…	…	…	Baseline

AUROC indicates area under the receiver operating characteristic; BP, blood pressure; EMR, eletronic medical record; ICC, intraclass correlation coefficient; PROMIS, Patient‐Reported Outcomes Measurement Information System; r, Pearson product moment correlation; and RICH LIFE, Reducing Inequities in Care of Hypertension: Lifestyle Improvement for Everyone.

^*According to Katzan and colleagues, ⁸⁶ ordinal α was used because it more accurately estimates reliability for data on a Likert scale as compared with the more widely used Cronbach α.

To examine baseline determinants of SDM, the outcome was SDM, and the predictors were the determinants of decision and included age, sex, race and ethnicity, education, health status, hypertension knowledge, attitudes and perceptions, patient activation, and health literacy, also adjusting for intervention group assignment.

To examine the relationship between SDM and change in BP among patients with hypertension, SDM, time, and SDM by time interaction were included in the model, adjusting for intervention group assignment.

SDM was assessed using CollaboRATE, ⁴⁷ a 3‐item patient‐reported measure containing the questions: “How much effort was made to help you understand your health issues?”, “How much effort was made to listen to the things that matter most to you about your health issues?”, and “How much effort was made to include what matters most to you in choosing what to do next?” At 12 months, participants were asked to rate each CollaboRATE item on a 10‐point scale (0=no effort was made, 9=every effort was made). The CollaboRATE mean sum score was used in the analysis, and higher scores indicated greater SDM. CollaboRATE was previously tested in both English and Spanish, in addition to being developed in collaboration with patients.

Systolic BP was measured by trained staff following a standardized measurement protocol using the OMRON 907XL BP device. An automated confirmatory BP measurement was to be performed if systolic BP was >140 mm Hg and/or the diastolic BP was ≥90 mm Hg. For data quality, in addition to training the staff of the clinics and using a standardized measurement protocol, the RICH LIFE project data collection started only after 3 months of device implementation.

The clinical characteristic included was health status assessed using the 10‐item 5‐point Likert scale Patient‐Reported Outcomes Measurement Information System global health measure. ⁴⁸ , ⁴⁹ , ⁵⁰ The Patient‐Reported Outcomes Measurement Information System includes items such as “In general, would you say your health is (5=excellent, 1=poor).” Scores of the Patient‐Reported Outcomes Measurement Information System global physical health are reported as T scores and are interpreted as poor (≤35), fair (35–42), good (42–50), very good (50–58), and excellent (>58). ⁵¹

In regard to hypertension knowledge and beliefs (perceptions and attitudes), hypertension knowledge was assessed with the item “What does the term hypertension mean?”, and responses were dichotomized (1=answered high BP, 0=all other answers: high level stress/tension, nervous condition, high blood sugar, overactivity). Hypertension perceptions was evaluated with the item “How serious of a personal health concern has high BP been?”, and responses were dichotomized (1=very serious, 0=all other answers: somewhat serious, and not at all serious). Hypertension attitudes was assessed with the item “How important do you think taking medicine is to keeping BP under control?”, and responses were also dichotomized (1=very important, 0=all other answers: somewhat important, and not at all important).

Personal resources to make and implement the decision included patient activation, and health literacy. Patient activation was assessed using the Patient Activation Measure, ⁵² a 13‐item 4‐point Likert scale (1=disagree strongly, 5=agree strongly) containing questions such as “Taking an active role in my own health care is the most important thing that affects my health.” Patient Activation Measure scores range from 0 to 100 and higher scores indicate higher patient activation.

Health literacy was measured using the Screening Questions for Limited Health Literacy, ⁵³ a 3‐item 5‐point Likert scale with questions such as: “How often do you have someone help you read instructions, pamphlets, or written materials from your doctor or pharmacy?” (0=none of the time, 4=all of the time). Health literacy was dichotomized as inadequate and adequate. Adequate was defined as selecting none or a little of the time for the first 2 questions, and extremely or quite a bit for the last question. Any other combination of answers was defined as inadequate health literacy.

Power Analysis

Power analysis calculation took into consideration unequal sample sizes. Therefore, instead of variance accounted for, calculations focused on how different the groups needed to be on SDM scores, based on standardized units, that is, Cohen d effect size. With a sample of 1426, we would be able to detect significant difference in SDM if the effect size was ≥0.29 when comparing Latino and non‐Latino White people and 0.16 for non‐Latino Black versus non‐Latino White people. Power was set at 0.80 and α at 0.05.

Statistical Analysis

Data were largely complete (<1% missingness), except for 12‐month systolic BP (15%). To properly address the missing data, we used a generalized estimating equations model. Income had a substantial amount of missing data (26.2%); to address this our analysis included “did not report” along with other income categories. Differences in SDM by race and ethnicity were evaluated using ANOVA for continuous variables and χ² test for categorical variables. We first conducted a linear regression analysis testing mean SDM and each determinant of decision variable (age, sex, race and ethnicity, education, income, employment, marital status, health status, patient activation, health literacy, knowledge that high BP is the same as hypertension, considering high BP as a serious health concern, and considering taking BP medication as very important) individually, controlling for group assignment. Next, we estimated a multivariable linear regression model with all significant determinants from the first analysis (bivariate) and known covariates of SDM (age and sex), ⁵⁴ controlling for group assignment. All assumptions for the multivariable linear regression model were evaluated. The residuals were homoscedastic and followed a normal distribution. Given that determinants of decision could be closely related, we assessed multicollinearity using the variance inflation factor. Results ranged from 1.04 to 2.34, indicating that multicollinearity was not found in the model. ⁵⁵ Lastly, we used generalized estimating equations to test if the change in systolic BP over time varied by level of SDM, controlling for covariates significant in the prior multivariable analysis and known covariates of SDM (age and sex). ⁵⁴ Analyses were conducted with the software Stata SE 17.

RESULTS

Participant Characteristics

At baseline, the sample (n=1426) had a mean age of 60±11.6 years (range, 22–96) and were mostly women (847, 59.4%). Participants consisted of non‐Latino Black (844, 59%), non‐Latino White (459, 32%), and Latino (123, 9%) people with uncontrolled BP. Demographic and other characteristics are stratified by racial and ethnic group in Table 2. Overall, the majority had an education level at high school or below (931, 65%), almost all participants had health insurance (1420, 99.8%), and health status was fair or poor (557, 39%). Higher score indicates greater SDM and, in our sample, the SDM mean score was 7.2±2.6 out of 9.

Table 2.

Determinants of Decision Measured at Baseline by Ethnic Group

Characteristic	NL White	NL Black	Latino	Total	P value*
Sample, n (%)	459 (32.2)	844 (59.2)	123 (8.6)	1426 (100)
Sociodemographics
Age, y, mean±SD	65±11.3	58±11.1	57±10.5	60±11.6	<0.001
Men, n (%)	224 (48.8)	310 (36.7)	45 (36.6)	579 (40.6)	<0.001
Education, n (%)					<0.001
<High school	22 (4.8)	170 (20.2)	65 (53.3)	257 (18.1)
High school	196 (42.8)	437 (51.8)	41 (33.6)	674 (47.4)
>High school	240 (52.4)	236 (28.0)	16 (13.1)	492 (34.6)
Annual household income, n (%)					<0.001
<$20 000	36 (9.5)	211 (36.2)	59 (64.8)	306 (29.1)
$20 000–$39 000	72 (19.0)	149 (25.6)	18 (19.8)	239 (22.7)
≥$40 000	271 (71.5)	223 (38.2)	14 (15.4)	508 (48.2)
Did not report	80 (17.4)	261 (30.9)	32 (26.0)	373 (26.2)
Employed full time and part time, n (%)	182 (39.6)	318 (37.9)	47 (38.2)	547 (38.5)	<0.001
Married, n (%)	262 (57.1)	249 (29.6)	52 (42.3)	563 (39.6)	<0.001
Have health insurance, n (%)	459 (100)	841 (99.8)	120 (98.0)	1420 (99.8)	0.009
Clinical
SBP, mm Hg, mean±SD	149.9±9.6	153.5±13.1	151.8±10.9	152.2±12.0	<0.001
DBP, mm Hg, mean±SD	81.0±11.3	87.6±12.1	85.4±11.0	85.3±12.1	<0.001
Health status, mean±SD†	46.2±8.6	44.1±8.9	41.8±9.2	44.6±8.9	<0.001
Poor, n (%)	49 (10.7)	132 (15.6)	30 (24.4)	211 (14.8)
Fair, n (%)	86 (18.7)	220 (26.1)	40 (32.5)	346 (24.3)
Good, n (%)	147 (32.0)	260 (30.8)	24 (19.5)	431 (30.2)
Very good/excellent, n (%)	177 (38.6)	232 (27.5)	29 (23.6)	438 (30.7)
Hypertension knowledge and beliefs
Knows that high BP means hypertension, n (%)	377 (82.1)	661 (78.3)	92 (74.8)	1130 (79.2)	0.119
High BP is a serious health concern, n (%)	196 (42.7)	579 (68.6)	94 (76.4)	869 (60.9)	<0.001
Taking BP medication is very important, n (%)	413 (90.0)	756 (89.7)	121 (98.4)	1290 (90.5)	0.008
Resources to make and implement decisions
Patient activation, mean±SD‡	66.5±15.1	66.8±16.6	57.6±13.5	65.9±16.1	0.004
Adequate health literacy, n (%)§	356 (77.6)	579 (68.6)	47 (38.2)	982 (68.9)	<0.001
Decision making
Shared decision making, median (IQR)‖	8 (5.3–9)	8.7 (7–9)	9 (6–9)	8.3 (6.3–9)	<0.001

BP indicates blood pressure; DBP, diastolic blood pressure; IQR, interquartile range; NL, non‐Latino; and SBP, systolic blood pressure.

^* P<0.05 comparison across ethnic groups.

^† Patient‐Reported Outcomes Measurement Information System (PROMIS) global physical health T scores ≤35: poor; 35–42: fair; 42–50: good; 50–58: very good; and >58: excellent.

^‡ Patient Activation Measure (PAM‐13), scores 0–100, higher scores indicate higher activation.

^§ Screening Questions for Limited Health Literacy.

^‖ CollaboRATE scores range from 0 to 9; a higher score indicates greater shared decision making.

Baseline Determinants of SDM

Table 3 displays results from the linear regression analysis examining the associations among baseline characteristics and SDM at 12 months. Our bivariate analyses found that after controlling for intervention group assignment, SDM was significantly associated with race and ethnicity, education, patient activation, and hypertension knowledge and beliefs. The same variables were also significantly associated with SDM in the multivariable analysis, with the exception of considering high BP as a serious health concern.

Table 3.

Baseline Determinants of Decision and Shared Decision Making at 12 Months (n=1426)

Variables (n)	Bivariate analysis		Multivariable analysis*
	β Coefficient	P value	β Coefficient	P value
Sociodemographic characteristics
Age	−0.02	0.365	0.02	0.587
Men	0.00	0.950	0.02	0.538
Race and ethnicity, NL White people (459)	Reference		Reference
NL Black people (844)	0.16†	<0.001†	0.14†	<0.001†
Latino people (123)	0.04	0.120	0.03	0.417
Education <high school	Reference		Reference
High school	−0.06	0.114	−0.05	0.178
>High school	−0.11†	0.003†	−0.08†	0.045†
Income <$20 000	Reference
$20 000–$39 000	−0.01	0.777
≥$40 000	−0.03	0.358
Did not report	0.03	0.458
Employed	−0.01	0.587
Married	−0.03	0.322
Clinical characteristics
Health status (mean)‡	0.03	0.350
Hypertension knowledge and beliefs
Knows that high BP is the same as hypertension	−0.06†	0.019†	−0.05†	0.046†
High BP is a serious health concern	0.07†	0.006†	0.03	0.245
Taking BP medication is very important	0.07†	0.009†	0.06†	0.022†
Resources to make and implement decisions
Patient activation (mean)§	0.08†	0.002†	0.09†	0.001†
Adequate health literacy‖	−0.01	0.618

BP indicates blood pressure; and NL, non‐Latino.

^*Adjusting for known covariates of shared decision making (age and sex) and for covariates that were significant in the bivariate analysis (race and ethnicity, education, knowledge that high BP is the same as hypertension, high BP is a serious health concern, taking BP medication is very important, and patient activation).

^† Statistically significant (P<0.05).

^‡ Patient‐Reported Outcomes Measurement Information System (PROMIS) global physical health.

^§ Patient Activation Measure (PAM‐13).

^‖Screening Questions for Limited Health Literacy.

When compared with non‐Latino White people, non‐Latino Black people had 0.14 points higher mean SDM score (P<0.001), and Latino people had 0.03 (P=0.417). We also found higher SDM mean scores for those who believe that taking BP medication is very important (β=0.06, P=0.022) and those with higher mean patient activation (β=0.09, P=0.001). On the contrary, participants with more than high school education (β=−0.08; P=0.045) and knowledge that high BP is the same as hypertension had lower mean SDM score (β=−0.05, P=0.046).

Systolic BP and SDM

Table 4 displays results of the association of SDM with systolic BP over 12 months. For adjusting this analysis, we included known covariates of SDM (age and sex) from previous literature ⁵⁴ and the variables that were significantly associated with SDM in the multivariable analysis (race and ethnicity, education, knowledge that high BP is the same as hypertension, belief that taking BP medication is very important, and patient activation). Our study found that, after controlling for intervention group assignment and the covariates mentioned, there was a greater reduction in systolic BP from baseline to 12 months with higher SDM mean scores (B=−0.42, P=0.035).

Table 4.

Association of SDM With Systolic BP Over 12 Months (n=1426)^*

Variables	β Coefficient	95% CI	P value
Time	−11.41	−14.41 to −8.40	<0.001
SDM†	0.15	−0.13 to 0.44	0.294
Time by shared decision making	−0.42	−0.81 to −0.03‡	0.035‡
Covariates
Age	0.12	0.07 to 0.18	<0.001
Men	−0.32	−1.54 to 0.89	0.604
Race and ethnicity NL White people	Reference
NL Black people	3.14	1.46 to 4.82	<0.001
Latino people	−0.21	−3.57 to 3.15	0.903
Education <high school	Reference
High school	−0.57	−2.34 to 1.2	0.527
>High school	−0.55	−2.62 to 1.53	0.607
Knows the meaning of hypertension	1.71	0.22 to 3.19	0.024
Taking BP medication is very important	−1.72	−3.75 to 0.3	0.096
Patient activation§	0.01	−0.02 to 0.05	0.442
Diabetes	1.60	0.39 to 2.82	0.010
Intervention	−4.80	−33.15 to 23.55	0.740

BP indicates blood pressure; NL, non‐Latino; and SDM, shared decision making.

^*Adjusted for known covariates of SDM (age, and sex), covariates that were significant in the multivariable analysis (race and ethnicity, education, knowledge that high BP is the same as hypertension, taking BP medication is very important, and patient activation), covariate that influence BP (diagnosis of diabetes), practice site, health system, and intervention group assignment.

^† SDM measured using mean CollaboRATE score.

^‡ Statistically significant (P<0.05).

^§ Patient Activation Measure (PAM‐13).

DISCUSSION

Main Findings

To our knowledge, this is the first study to examine the relationships among SDM, BP, and race and ethnicity among patients with hypertension. Our study revealed some key findings: (1) Black people had higher levels of self‐reported SDM than White people. (2) Patient activation and believing that taking BP medication is very important were positively associated with SDM. (3) Patients with some education beyond high school reported lower levels of SDM than those with less than high school education. (4) Knowledge that high BP is the same as hypertension was negatively associated with SDM. (5) There was a greater reduction in systolic BP over time among patients with a higher SDM mean score at 12 months. Our hypotheses that SDM scores would differ by patients' race and ethnicity and that higher SDM scores would be associated with greater reduction in BP were supported.

Association With Determinants of Decision Including Race and Ethnicity

Our study found differences in SDM scores by race and ethnicity and education. Non‐Latino Black people had higher SDM scores than non‐Latino White people. Forcino and colleagues analyzed 31 265 survey responses and found that SDM scores were significantly associated with race and education. ⁵⁴ Cooper‐Patrick and colleagues reported that Black people perceived their medical visits as significantly less participatory compared with White individuals. ⁵⁶ However, they also found that patients experienced greater participation when they had clinicians of the same race, in contrast to those with clinicians of a different race. Corroborating these findings, in our recently published mixed‐methods article, participants described mutual understanding as a facilitator of SDM in racial concordant visits. ³⁶ Although our study lacks clinician data, race concordance remains a plausible explanation for our findings and warrants further investigation. Additional research into SDM experiences across diverse patient and provider racial identities may provide greater insight into these relationships.

Among Latino participants, the SDM scores were not statistically different from those of non‐Latino White participants. In agreement, Ratanawongsa et al showed that in a national sample of adult US Latino people, they self‐reported no differences in SDM in the context of hypertension. ²⁹ Other studies have also shown that Latino patients were more likely to allow their physicians to take a predominant role in making decisions related to their health, ⁵⁷ and that they also expressed preferences for a more passive role in these decisions. ⁵⁸ In alignment with these studies and given that SDM was self‐reported in our study, a plausible explanation for our findings can be that Latino participants could have had a lower expectation of what SDM entails and scored it higher than groups. Patients from diverse racial and ethnic and cultural backgrounds differ when it comes to advocating for their health care and appraising treatment‐related decision making, ¹⁶ highlighting the need to further explore SDM in diverse samples and using other data collection approaches such as direct observation.

Additionally, the broad categorization of all Latino populations in 1 unique group may mask important variations within the various Latino subgroups in terms of cardiovascular health and SDM, conferring the need to explore SDM using a large sample size of Latinos that allows for within‐group comparisons. ⁵⁹

Surprisingly, those with greater than high school education reported lower SDM scores than those with less than high school education. Additionally, those with knowledge that high BP is the same as hypertension also reported lower SDM scores. We hypothesized that the need for SDM might be perceived as less critical in this context, because the health care providers might presume that these individuals are better able to comprehend and are already engaged in informed discussions. Our results corroborate with those from Forcino and colleagues on education as well. In their study, when compared with participants with eighth grade educational attainment or less, 4‐year college graduates were 36% less likely to rate SDM with the highest score (odds ratio [OR], 0.638 [95% CI, 0.520–0.783]; P<0.001) and those with higher than a 4‐year college degree were 39% less likely (OR, 0.613 [95% CI, 0.500–0.752]; P<0.001). ⁵⁴ Durand and colleagues reported on a meta‐analysis showing a positive effect of SDM interventions among disadvantaged patients. ⁶⁰ The interventions appeared to benefit disadvantaged groups more than those with higher literacy, education, and socioeconomic status. Furthermore, Peek and colleagues performed an observational cohort study using a comprehensive patient education on SDM among Black people with diabetes. ¹⁶ Their findings shed light on improvements in various outcomes, including hemoglobin A1c, high‐density lipoprotein cholesterol, self‐efficacy, and self‐care behaviors. These results suggest that SDM interventions could be tailored to meet the needs of specific populations to narrow health disparities.

Participants with higher patient activation and those who considered taking BP medication as very important reported higher SDM scores. A study by Poon et al analyzed survey responses of 1222 adult patients with diabetes and/or cardiovascular disease. ⁶¹ They reported bidirectionality in the relationship between patient activation and patients' experiences of SDM. An article by Smith and colleagues reporting nationally representative data found that higher patient activation was associated with greater perceived benefit of SDM across various types of health care‐related decisions. ⁶² Other studies also reported that SDM improves adherence to medication and medication management. ⁶³ , ⁶⁴

Reduction in Systolic BP

Our study indicated that a reduction of ≥10 mm Hg was observed when a higher SDM was experienced, suggesting a benefit of promoting SDM in hypertension care. ³⁶ Such a reduction is considered clinically significant, because it can substantially lower the risk of cardiovascular events such as coronary heart disease, stroke, and heart failure, thereby improving overall health outcomes in patients with hypertension and reducing mortality. ⁶⁵ In alignment with our findings, Kask‐Flight and colleagues ⁶⁶ published results of a cluster randomized clinical trial with 130 male participants and reported significant reductions in BP among patients with stage 2 hypertension (mean difference of change: −7.86, P=0.038). The intervention used an interactive computer‐based decision aid, consisting of individualized cardiovascular risk calculations, visual figures of risk summaries, and suggestions for lifestyle interventions. In addition, providers underwent 4 hours of training on the use of the decision aid and on SDM processes.

In contrast to our study's findings, Johnson and colleagues ³¹ published a systematic review in which 6 controlled trials evaluated the effects of SDM interventions for adults with hypertension. Included studies with usable data evaluating BP outcome showed nonsignificant reduction in BP. ⁶⁷ Potential explanations for these null findings could be related to the studies' high risk of bias. ³¹ Noteworthy, none of the interventions used in these studies were developed based on established guidelines, ⁶⁸ they were heterogeneous in their content, generally used various components, the focus on providers training was minimal, and control groups consisted of different comparators that could have contaminated the control sample. In addition, a more recent randomized comparative effectiveness trial publication by Boulware and colleagues ⁶⁷ also found no group differences in BP outcomes after a SDM intervention for Black people. The SDM intervention consisted of a 1‐time 30‐minute training for patients based on 4 key PART steps (P for prepare for visits, A for act during the visits, R for review key recommendations, and T for take recommendations home). In further exploration of the trial's protocol ⁶⁹ to understand what the PART steps entailed, the intervention is described as including only 1 of the components of SDM, patients' preferences, and therefore cannot be considered an SDM intervention. The article by Boulware and colleagues ⁶⁷ highlights the misinterpretation of SDM as an additional obstacle to fully understanding the role of SDM in hypertension management.

Studies evaluating the impact of SDM interventions on BP control are scarce and inconsistent. The few studies published to date are clinically and methodologically heterogenous, with noted variations in the interventions, comparators, and outcomes. ³¹ Therefore, current evidence is insufficient to evaluate the effectiveness of existing SDM interventions in the context of hypertension care and to inform which intervention should be used.

Implications for Practice, Research, Education, and Policy

Patient participation in decisions involving their health is regarded not only as ethical, but also as an important aspect of high‐quality clinical practice, health care systems, public health, and health care policy. ⁸ , ³² , ⁷⁰ , ⁷¹ , ⁷² , ⁷³ , ⁷⁴ Provisions of the US Affordable Care Act promote or require SDM. ⁷⁰ , ⁷⁵ SDM was stipulated by the Centers for Medicare and Medicaid Services as a prerequisite to reimbursement for certain health conditions. ⁷⁶ The American College of Cardiology, American Heart Association, American Medical Association, and National Quality Forum also endorse SDM and its potential to significantly transform medical culture and health systems. ⁷⁷ , ⁷⁸ , ⁷⁹ , ⁸⁰ These legislative and guideline efforts require a robust evidence base on the effectiveness of SDM in clinical settings. However, the current evidence to guide the use of SDM in hypertension management is limited.

Our study findings, added to current evidence, collectively emphasize the importance of SDM in clinical practice to achieve better hypertension outcomes, including for people with lower literacy, education, and socioeconomic status. SDM should therefore be incorporated in clinical practice so that care can be tailored to patients' needs and priorities. ⁶¹ Individuals present various levels of patient readiness or activation, therefore requiring different types of support in clinical practice to engage in SDM. ⁶¹ Our results also show an association between SDM and patients' perceived importance of taking medication, further highlighting the opportunity to use SDM strategies to promote medication adherence to improve hypertension control. ⁹

Our study results add to the previously limited evidence on SDM in the context of hypertension care ³¹ and support the need for future studies with robust methodology comparing various types of SDM interventions. Studies should use or develop SDM interventions that follow established SDM definitions and components, such as those recommended by Bomhof‐Roordink and colleagues ¹² and Stacey and colleagues ⁸¹ through the IPDAS (International Patient Decision Aid Standards) collaboration. Such studies have the potential to improve present knowledge on interventions that are most effective, including those targeting providers. Given the influence of culture in SDM, ³⁹ future research with racially and ethnically diverse samples and in a wide array of countries could further expand our understanding of SDM benefits in hypertension treatment and management across populations. Patients from diverse racial and ethnic and cultural backgrounds differ when it comes to advocating for their health care and appraising treatment‐related decision making, highlighting the need to further explore SDM in diverse samples. ¹⁶ Furthermore, future research should examine ways in which SDM can be responsive to social determinants of health that influence health outcomes to address important root causes of hypertension disparities. ²¹ , ⁸²

Results from this study indicating that non‐Latino Black people had higher SDM scores and that individuals with higher education had lower SDM scores challenge some preconceived ideas about SDM. A systematic review by FitzGerald and Hurst reported that diagnosis, treatment decisions, and levels of care are likely influenced by health care professionals' implicit biases. ⁸³ Therefore, SDM curricular and training efforts for health care professionals should include discussions on implicit bias. These discussions can promote awareness of these issues and consequently support providers in the implementation of SDM by eliciting patients' values and preferences despite any preconceived ideas. Moreover, according to Thériault and colleagues, SDM needs to be integrated across the entire curriculum rather than being isolated in stand‐alone courses. ⁸⁴ Educational efforts would be more beneficial if the context in which SDM is taught includes patient representation from various backgrounds and contextual issues related to social determinants of health.

Our results show variation in SDM among patients from different racial and ethnic groups. Therefore, policy should reflect these differences and recommend patient decision aids ⁶⁸ and other interventions to facilitate SDM that have been tested and proven effective for various racial and ethnic groups. Furthermore, given the incipient state of SDM in hypertension research body, ³¹ policy makers should make sure that promotion of SDM implementation in clinical practice takes these limitations into consideration by providing definitions for desired outcomes, offering guidance on measurement tools to be used, evaluating the impact of policies over time with reevaluation at regular intervals, ⁸⁵ and updating policies as SDM in hypertension research evolves.

Study Limitations and Strengths

We also acknowledge the limitations of our study. Our observational design prevents establishment of causality. SDM was measured only at 12 months, precluding an exploration of SDM change over time. Data on SDM were self‐reported, which is subject to acquiescence and social desirability bias. Limitations were attenuated by the use of data from a pragmatic trial with robust execution and methodology, by using an SDM measurement tool that was previously tested in both English or Spanish, in addition to being developed in collaboration with patients. ⁴⁷

Although this study included comparisons among racial and ethnic groups, other growing populations that may experience a high burden of hypertension were not represented. These include groups for whom SDM might be particularly important, such as American Indian and Alaska Native, Native Hawaiian and Other Pacific Islander, and Asian populations. Moreover, a potential limitation of our data source relates to the inability to further categorize participants in racial and ethnic subgroups.

CONCLUSIONS

We found that SDM is positively associated with BP reduction, and determinants of decision, including race and ethnicity, education, hypertension knowledge and attitude, and patient activation are associated with SDM at follow‐up. Future research should further explore SDM‐related differences among racial and ethnic groups to better align care with the needs of patients with hypertension. Our results offer actionable information to enhance clinical practice and policy improvement and to support the development of effective clinician and patient training strategies and decision aids to support patient engagement in decisions about their hypertension care.

Sources of Funding

The RICH LIFE project was supported by a grant from the National Institutes of Health (NIH) National Heart, Lung, and Blood Institute (NHLBI) through a partnership with the Patient‐Centered Outcomes Research Institute (UH2/UH3 HL130688). Dr Elias was supported by a grant from the National Institutes of Health (NIH) National Institute of Nursing Research (NINR) (F31 NR019523).

Disclosures

None.