Efficacy of a Discharge Educational Strategy vs Standard Discharge Care on Reduction of Vascular Risk in Patients With Stroke and Transient Ischemic Attack

This randomized clinical trial determines the efficacy of a culturally tailored skills-based educational intervention with telephone follow-up compared with standard discharge care on systolic blood pressure reduction in a multiethnic cohort of patients with mild/moderate stroke/transient ischemic attack.

Key Points

Question

What is the effect of a skills-based discharge intervention on blood pressure reduction among patients with mild/moderate stroke and transient ischemic attack?

Findings

This randomized clinical trial of 552 patients found that a culturally tailored, skills-based discharge program with follow-up reinforcement led to a nonsignificant 2.5 mm Hg–greater systolic blood pressure reduction at 12 months postdischarge compared with usual care. There was a statistically significant 9.9 mm Hg–greater systolic blood pressure reduction in the intervention group compared with usual care among Hispanic individuals.

Meaning

Further investigation may identify whether culturally tailored, skills-based education improves long-term vascular risk reduction for individuals who have had mild/moderate stroke/transient ischemic attack in different populations.

Abstract

Importance

Despite secondary prevention strategies with proven efficacy, recurrent stroke rates remain high, particularly in racial/ethnic minority populations who are disproportionately affected by stroke.

Objective

To determine the efficacy of a culturally tailored skills-based educational intervention with telephone follow-up compared with standard discharge care on systolic blood pressure reduction in a multiethnic cohort of patients with mild/moderate stroke/transient ischemic attack.

Design, Setting, and Participants

Randomized clinical trial with 1-year follow-up. Participants were white, black, and Hispanic patients with mild/moderate stroke/transient ischemic attack prospectively enrolled from 4 New York City, New York, medical centers during hospitalization or emergency department visit between August 2012 and May 2016. Through screening of stroke admissions and emergency department notifications, 1083 eligible patients were identified, of whom 256 declined to participate and 275 were excluded for other reasons. Analyses were intention to treat.

Interventions

The Discharge Educational Strategies for Reduction of Vascular Events (DESERVE) intervention is a skills-based, culturally tailored discharge program with follow-up calls delivered by a community health coordinator. This intervention was developed using a community engagement approach.

Main Outcomes and Measures

The primary outcome was systolic blood pressure reduction at 12 months postdischarge.

Results

A total of 552 participants were randomized to receive intervention or usual care (281 women [51%]; mean [SD] age, 64.61 [2.9] years; 180 Hispanic [33%], 151 non-Hispanic white [27%], and 183 non-Hispanic black [33%]). At 1-year follow-up, no significant difference in systolic blood pressure reduction was observed between intervention and usual care groups (β = 2.5 mm Hg; 95% CI, −1.9 to 6.9). Although not powered for subgroup analysis, we found that among Hispanic individuals, the intervention arm had a clinically and statically significant 9.9 mm Hg–greater mean systolic blood pressure reduction compared with usual care (95% CI, 1.8-18.0). There were no significant differences between arms among non-Hispanic white (β = 3.3; 95% CI, −4.1 to 10.7) and non-Hispanic black participants (β = −1.6; 95% CI, −10.1 to 6.8).

Conclusions and Relevance

Few behavioral intervention studies in individuals who have had stroke have reported clinically meaningful reductions in blood pressure at 12 months, and fewer have focused on a skills-based approach. Results of secondary analyses suggest that culturally tailored, skills-based strategies may be an important alternative to knowledge-focused approaches in achieving sustained vascular risk reduction and addressing racial/ethnic stroke disparities; however, these findings should be tested in future studies.

Trial Registration

ClinicalTrials.gov identifier: NCT01836354.

Introduction

Approximately 200 000 of the 800 000 stroke events occurring annually in the United States are recurrent events.¹ Compared with incident strokes, recurrent events are disproportionately fatal, disabling, and impose a high financial burden on patients, families, and the health care system.^2,3,4 Individuals who have had an incident stroke have up to a 30% probability of having another event within 5 years,⁵ and rates of recurrence are significant even among those with mild stroke and transient ischemic attack (TIA) who also have significant vascular risk factor profiles.^6,7 Recurrent events disproportionately affect African American and Hispanic individuals; they are 2 to 3 times more likely to experience stroke and experience higher stroke-related mortality compared with non-Hispanic white individuals.^8,9,10,11,12

While clinical evidence strongly supports the effectiveness of improving health behaviors (eg, medication adherence) and modifying vascular risk factors (eg, blood pressure [BP] reduction) to reduce recurrent events,¹³ secondary stroke prevention remains elusive. Within 1 year of hospital discharge, up to one-third of patients are not adherent to prescribed medications.¹⁴ Patients with mild stroke and TIA, in particular, may underestimate risk of recurrence owing to mild or resolved deficits.¹⁵ There are also persistent racial/ethnic disparities in secondary prevention, largely due to cultural and behavioral factors not addressed in standard prevention modalities and poor integration into community resources.^11,16,17,18

A number of previous structural and educational interventions have targeted BP control in secondary stroke prevention. Structural interventions have focused on implementation of health system strategies to ensure patients are discharged appropriately and receive adequate follow-up care^19,20; however, most tested interventions have not effectively addressed issues relevant to low socioeconomic and diverse populations. Previous work strongly suggests the context of behavior change plays an important role in development of optimal prevention strategies.^6,17,21 To this end, various trials have implemented educational interventions aimed at risk factor reduction with mixed results.^22,23 Educational interventions to date have focused on teaching patients about their risk but have not coupled that with self-efficacy from a skills-based approach. Further, few of these interventions have embraced culturally tailored strategies, which are critically needed to respond to health disparities.

Discharge Educational Strategies for Reduction of Vascular Events (DESERVE) is an interactive, skills-based behavioral intervention delivered by a community health coordinator at hospital discharge, with 3 postdischarge reinforcement calls. DESERVE was a mandate from our stroke community as part of continued community engagement from a previous discharge stroke preparedness trial, the Stroke Warning Information and Faster Treatment (SWIFT) study.⁶ Participants and family members in the SWIFT study requested skills and action steps to reduce risk; participants contributed to developing DESERVE intervention materials and are featured in patient narrative videos. The present study was a randomized clinical trial (RCT) that assessed the efficacy of DESERVE in a multiethnic cohort of patients with mild/moderate stroke/TIA. The primary outcome was difference in systolic BP (SBP) reduction between predischarge and 12 months postdischarge.

Methods

Study Design, Setting, and Participants

DESERVE was a 2-arm RCT with a 6- and 12-month follow-up. Between August 2012 and May 2016, 552 patients were prospectively enrolled from New York University Langone Medical Center, Icahn School of Medicine at Mount Sinai, Columbia University Medical Center, and Bellevue Hospital Center in New York City. A detailed description of the screening process and eligibility criteria has been published previously.²⁴ The trial protocol is available in the Supplement. All participants provided written informed consent. The study was approved by the institutional review boards of participating medical centers. Analyses began in May 2017.

Intervention

Participants were randomized, stratified by language and site, into intervention or usual care groups in a 1:1 ratio using a computerized permuted block randomization program. Prior to discharge, the intervention group engaged in an interactive educational session with a community health coordinator, someone hired from the underlying patient population and trained to do research; this position is similar to a community health worker or navigator. Participants also received a patient-paced workbook and video emphasizing 3 skills-based thematic areas: patient-physician communication, medication adherence, and accurate stroke risk perception with risk reduction skills (Table 1). Details of intervention development and materials are described elsewhere.²⁴ The intervention group also received follow-up calls from a coordinator at 72 hours, 1 month (typically immediately prior to neurology appointment), and 3 months to enhance strategies for patient-physician communication, clarify medication regimens, and provide social support to motivate behavior change. Participants randomized to the usual care group received standard stroke treatment as well as American Heart Association stroke pamphlets in English or Spanish.

Table 1. Intervention Summary.

Group	Baseline	Discharge	72 h	1 mo	3 mo	6 mo	12 mo
Intervention	Interview	Educational PowerPoint presentation; motivational video; delivery of workbook	Poststroke contact form; verify follow-up physician visit	1-mo Follow-up form; physician visit reminder; review patient-physician communication in workbook; follow-up call after physician visit	3-mo Follow-up form; follow-up for other physician visits; review main workbook components; discuss other resources	In-person visit/form; blood pressure, hemoglobin A1c, anthropometric measurements	In-person visit/form; blood pressure, hemoglobin A1c, anthropometric measurements
Usual care	Interview	Standard discharge education	Patient-guided contact	Patient-guided contact	Patient-guided contact	In-person visit/form; blood pressure, hemoglobin A1c, anthropometric measurements	In-person visit/form; blood pressure, hemoglobin A1c, anthropometric measurements

Study Procedures and Outcomes

Validated measures including vascular risk metrics (Table 1) were used at baseline and at 6 and 12 months postdischarge.²⁴ Race and Hispanic/Latino ethnicity were self-reported by participants by asking them to select from a list of US Census categories. Follow-up surveys and measures were conducted by trained research assistants blinded to intervention status. Every attempt was made by the study team to conduct follow-up surveys in person. If unable to complete in person, participants were interviewed over the telephone, and permission was obtained to request recent laboratory work from the participant’s physician. We reviewed records for BP as well as secondary outcomes, including other risk factors and outcome events of stroke, myocardial infarction, or vascular death; adjudication of secondary outcomes is currently under way. If participants could not be reached by telephone, retention letters were sent and home visits were made.

The primary study outcome was SBP reduction between baseline (predischarge) and 12 months postdischarge. Baseline BP was measured at least 48 hours poststroke up to 3 times by trained coordinators or research assistants using an automatic sphygmomanometer, following American Heart Association guidelines.²⁵ At 12-month follow-up, 120 BP measurements (25%) were done in person in the study office or participants’ homes, 294 (62%) came from either clinic visits where our study staff met patients or abstracted medical records, 26 (5%) were self-reported, and 38 (8%) did not have a source recorded. The distribution of source of follow-up BP measurement did not differ significantly between trial arms. If participants could not be reached, we used the National Death Index to ascertain deaths and obtain cause of death; all other participants were considered lost to follow-up.

Statistical Analysis

Sample size calculations were based on a randomized pilot study of 50 patients with TIA/mild stroke, which demonstrated an 8–mm Hg reduction in SBP in the intervention vs usual care group. We then conducted power analysis using a lower 6–mm Hg estimate and an assumed variability of 23 mm Hg as seen in the pilot. Given 300 participants per group and α = .05, power was 89.1%. Mean reduction in SBP in the intent-to-treat population was compared between trial arms using linear regression models with 95% CI adjusted for study site and mixed-effects models using BP data at baseline and at 6- and 12-month follow-up. Secondary analyses were conducted to assess the effect of the intervention by race/ethnicity by stratifying all analyses by racial/ethnic group (non-Hispanic white, non-Hispanic black, and Hispanic). Non-Hispanic other was not included because it comprised a small and heterogeneous sample. In an additional exploratory analysis, the effect of the intervention by sex was evaluated. Although the study was not powered to detect differences in subgroup intervention effects, adjusted linear models with an interaction term were used to test statistical interaction between trial arm and race/ethnicity and between trial arm and sex. Within the intervention group, we also examined whether there was a dose-response relationship by number of follow-up reinforcement calls completed for the intervention response by comparing mean reduction in SBP by number of calls completed using t tests and analysis of variance. All analyses used SAS version 9.4 (SAS Institute), and statistical significance was determined where P was less than .05. All CIs are presented as nominal, ie, without adjustment for the multiplicity of tests performed.

Results

Recruitment and Follow-up

A total of 4166 participants were screened and 1083 (26%) were eligible for enrollment (Figure 1). We screened a very large population owing to nonspecific symptoms that may be associated with mild stroke/TIA. We enrolled 552 patients, with 274 (49.6%) randomized to intervention and 278 (50.4%) to usual care. At 12-month follow-up, 13 participants (4.7%) in the intervention group and 14 (5.0%) in the usual care group were alive but unable to be reached to determine SBP levels. Median (interquartile range) duration of follow-up to measure SBP reduction in the study (not shown) was 13.8 (12.2-17.4) months and did not differ between intervention (13.8 [12.3-17.1]) and usual care (13.7 [12.2-17.6]).

Figure 1. Discharge Educational Strategies for Reduction of Vascular Events Study Flow Diagram.

BP indicates blood pressure.

Patient Characteristics

Baseline characteristics were well balanced between trial arms overall, although participants in the intervention arm were slightly younger than in usual care (mean [SD] age, 63.3 [13.1] vs 65.9 [12.6] years, P = .02) (Table 2). Mean National Institutes of Health Stroke Scale score indicated that strokes in this patient sample were generally mild. In both trial arms, approximately half of participants were women and born in the United States, and slightly less than half had a high school education or less and were married or living with a partner. Participants in both arms were roughly equally distributed across racial/ethnic groups. Approximately one-third of participants had a prior stroke or TIA, and most had a history of hypertension diagnosis and had overweight or obesity.

Table 2. Baseline Demographic and Clinical Characteristics of Study Participants^a.

Characteristic	No. (%)		P Value
	Intervention (n = 274)	Usual Care (n = 278)
Women	139 (50.7)	142 (51.3)	.90
Race/ethnicityb
Non-Hispanic white	67 (24.8)	84 (30.8)	.46
Non-Hispanic black	96 (35.6)	87 (31.9)
Hispanic	93 (34.4)	87 (31.9)
≤High school education	125 (46.8)	130 (48.1)	.76
Married or living with partner	121 (45.8)	123 (45.4)	.63
Medicaid insurance coverage	93 (34.1)	93 (33.6)	.90
Born in the United States	153 (57.5)	162 (59.6)	.63
Smoking status
Current	40 (15.0)	42 (15.3)	.38
Former	84 (31.6)	72 (26.3)
Any physical activity for ≥10 min in past 30 d	131 (49.6)	122 (45.4)	.32
History of hypertension diagnosis	229 (86.4)	228 (85.7)	.82
Blood pressure predischarge
Normal	44 (16.1)	43 (15.5)	.57
Elevated	38 (13.9)	29 (10.4)
Stage 1c	62 (22.6)	72 (25.9)
Stage 2d	130 (47.5)	134 (48.2)
History of stroke or TIA	79 (30.6)	88 (33.8)	.43
Overweight or obesee	201 (73.4)	190 (68.4)	.20
Age, mean (SD), y	63.3 (13.1)	65.9 (12.6)	.02
Blood pressure, mean (SD), mm Hg
Systolic	139.4 (21.6)	139.8 (22.3)	.81
Diastolic	76.9 (13.0)	76.5 (14.1)	.73
Modified Rankin score, mean (SD)	1.4 (1.5)	1.5 (1.4)	.34
NIHSS score, mean (SD)
At admission	2.9 (3.3)	3.4 (4.0)	.12
At discharge	1.8 (2.6)	2.1 (3.1)	.22
Average h of sleep per night in past mo, mean (SD)	6.5 (1.9)	6.2 (1.7)	.09
Total cholesterol, mean (SD), mg/dL	174.2 (46.4)	176.0 (51.4)	.67

Abbreviations: NIHSS, National Institutes of Health Stroke Scale; TIA, transient ischemic attack.

SI conversion factor: To convert cholesterol to millimoles per liter, multiply by 0.0259.

^aNumbers may not sum to the total owing to missing observations.

^bEighteen intervention and 20 usual care participants were either non-Hispanic other race/ethnicity (reported being non-Hispanic and neither white or black) or did not report race or ethnicity.

^cStage 1 hypertension was defined as having systolic blood pressure 130 mm Hg to 139 mm Hg or diastolic blood pressure 80 mm Hg to 89 mm Hg.

^dStage 2 hypertension was defined as having systolic blood pressure 140 mm Hg or greater or diastolic blood pressure 90 mm Hg or greater.

^eBody mass index (calculated as weight in kilograms divided by height in meters squared) greater than or equal to 25.

SBP Reduction Between Baseline and 12-Month Follow-up

Mean (SD) reduction in SBP between baseline and 12-month follow-up was 7.0 (25.5) mm Hg among intervention and 4.3 (23.4) mm Hg among usual care participants (Table 3 and Figure 2). In adjusted models, no significant difference in reduction in SBP was observed between groups (β = 2.5 mm Hg; 95% CI, −1.9 to 6.9). However, the intervention arm had an almost 10 mm Hg–greater mean reduction in SBP compared with usual care among Hispanic participants (β = 9.9; 95% CI, 1.8-18.0). An additional analysis of the primary end point, using a random-effects model that also included the 6-month SBP assessment, also showed no overall difference in SBP reduction but a significant effect for the intervention among Hispanic individuals (estimated mean difference at 12 months = 9.4 mm Hg; 95% CI, 1.7-17.1). There were no significant differences in mean reduction of SBP between trial arms among non-Hispanic white individuals (β = 3.3; 95% CI, −4.1 to 10.7) and non-Hispanic black participants (β = −1.6; 95% CI, −10.1 to 6.8). The interaction term between intervention arm and race/ethnicity was not significant (F = 1.84; P = .16). Among men, the intervention arm had 5.5 mm Hg–greater mean reduction in SBP (95% CI, −0.7 to 11.8) compared with the usual care arm, but this difference was not statistically significant. There was no difference noted among women (β = –0.4; 95% CI, −6.5 to 5.8). The interaction term for intervention arm and sex was not significant (t = 1.37; P = .17). To address the statistically significant difference in mean age between trial arms at baseline, we conducted sensitivity analyses to also adjust for age in linear regression models and found similar results.

Table 3. Comparison of Systolic Blood Pressure Reduction From Baseline to 12-Month Follow-up Between Trial Arms.

Characteristic	Systolic Blood Pressure, Mean (SD), mm Hg				Systolic Blood Pressure Reduction, Baseline to 12-mo Follow-up, Mean (SD), mm Hg
	Intervention		Usual Care		All Participants	Intervention	Usual Care	Difference: Intervention vs Usual Care (95% CI)a
	Baseline	12 mo	Baseline	12 mo
All participants	139.4 (21.6)	133.1 (19.8)	139.8 (22.3)	135.2 (18.3)	5.6 (24.5)	7.0 (25.5)	4.3 (23.4)	2.5 (−1.9 to 6.9)
Race/ethnicityb
Non-Hispanic white	141.4 (22.0)	132.4 (17.7)	138.4 (18.8)	130.7 (16.8)	7.9 (21.6)	9.6 (22.1)	6.6 (21.2)	3.3 (−4.1 to 10.7)
Non-Hispanic black	139.5 (19.7)	135.4 (21.7)	144.6 (23.1)	138.6 (19.0)	6.2 (26.4)	5.8 (27.1)	6.7 (25.7)	−1.6 (−10.1 to 6.8)
Hispanic	139.2 (23.4)	131.6 (20.3)	136.8 (25.2)	138.1 (18.7)	3.0 (25.6)	7.7 (26.8)	−2.1 (23.3)	9.9 (1.8 to 18.0)
Sexb
Male	142.5 (23.0)	132.4 (21.0)	140.8 (19.5)	135.2 (17.6)	8.3 (24.5)	11.2 (27.5)	5.5 (20.9)	5.5 (−0.7 to 11.8)
Female	136.4 (19.6)	133.8 (18.7)	138.8 (24.8)	134.9 (18.8)	3.1 (24.2)	3.0 (22.9)	3.2 (25.7)	−0.4 (−6.5 to 5.8)

^aAdjusted for study site; interaction between trial arm and race/ethnicity, P = .16; interaction between trial arm and sex, P = .17.

^bDifference in mean reduction by race/ethnicity in overall sample, P = .22; by sex, P = .02.

Figure 2. Comparison of Systolic Blood Pressure Reduction Between Trial Arms.

^aSignificant at the .05 level.

Finally, we examined whether there was a dose-response relationship in the intervention group between participating in up to 3 follow-up calls and SBP reduction. Of those randomized to the intervention arm, two-thirds (184 [67.2%]) had 3 follow-up calls, 15.0% (n = 41) had 2 follow-up calls, 6.2% (n = 17) had 1 follow-up call, and 11.7% (n = 32) received no follow-up calls. We found that with each additional follow-up call, there appeared to be a greater mean reduction in SBP (1 call, 0.4 mm Hg; 2 calls, 6.1 mm Hg; 3 calls, 8.3 mm Hg; F = 0.77; P = .51). Also, those who received the full intervention (all 3 calls) had a greater reduction compared with those who received partial intervention (<3 calls) (8.3 vs 3.5 mm Hg, t = −1.32; P = .19). However, these differences were not statistically significant.

Discussion

The DESERVE trial focused on a novel skills-based approach to vascular risk reduction and is also one of the first stroke prevention RCTs to examine differences in behaviors and outcomes by race/ethnicity. At 12-month follow-up, the intervention group had a nonsignificant 2.5 mm Hg–greater mean SBP reduction compared with usual care (95% CI, −1.9 to 6.9). Although not powered to detect differences in subgroup analysis, we found that Hispanic participants in the intervention group had a clinically and statistically significant 9.9 mm Hg–greater SBP reduction compared with usual care (95% CI, 1.8-18.0). The SBP reduction in Hispanic individuals translates to nearly 40% risk reduction for secondary stroke events.²⁶ Hispanic individuals are the fastest growing population group in the United States²⁷ and have higher rates of incident stroke than non-Hispanic white individuals.^10,28 This results in a major burden on the health care system, with 1 estimate projecting the cost of ischemic stroke for Hispanic individuals from 2005 to 2050 to be $313 billion, in 2005 US dollars.²⁹ Culturally tailored, skills-based interventions such as DESERVE could lead to major health care savings as well as significant reductions in racial/ethnic disparities in stroke incidence and recurrence.

There appears to be a different pattern of response to the intervention and usual care across racial/ethnic groups (Table 3). Among Hispanic individuals, there was a large reduction in SBP in the intervention group and a small increase in usual care. Similar to Hispanic individuals, non-Hispanic white individuals experienced a large reduction in SBP in the intervention group, but they also had a moderate reduction in BP in usual care. The pattern is different in non-Hispanic black individuals, with the usual care and intervention groups both experiencing moderate decreases in BP. We speculate this differential response may be due to a number of factors, including differential uptake or use of the intervention, different social network structures, and poorer access to stroke resources among Hispanic individuals.³⁰

Hispanic individuals in the United States have low stroke-specific health literacy and are less likely to be aware of their hypertension status, adhere to cardiovascular disease medications, and have controlled hypertension compared with non-Hispanic white individuals.^{31,32,33,34,35,36,37,38} The effect of the DESERVE intervention on Hispanic individuals may be due to specific components of the intervention design that address some of these gaps. The intervention included English and Spanish materials and bilingual coordinators delivering the intervention who were ethnically concordant with participants whenever possible. Among Hispanic individuals, our culturally tailored stroke survivorship narrative video developed in partnership with community members framed recovery in the context of faith and spirituality, in contrast to the video for African American individuals, which presented recovery after stroke as a matter of self-determination. Importantly, DESERVE educational materials also focused on building patient-physician communication skills, which are a documented barrier to care among Hispanic individuals.^{39,40,41,42,43}

Previously, several structural interventions have tested models of collaborative and coordinated care for optimizing secondary stroke prevention.^19,20,44,45 While some proved effective, they were mainly conducted in ethnically homogenous populations and within single-payer health care systems; it is unclear whether such approaches would be feasible in the United States where health care differentials exist. In contrast to these structural approaches to secondary prevention, other strategies focused on improving patients’ knowledge through education. These interventions have taken various forms, including education at discharge alone, education with reinforcement postdischarge, and ongoing BP monitoring.^22,23,46,47 Most tested educational interventions use highly trained staff, such as nurses, and involve extensive time and effort, a model that is not sustainable given the high cost of health care.

DESERVE builds on past interventions in a number of significant ways. While previous educational interventions have largely focused on changing patients’ risk factor knowledge, DESERVE is one of the first interventions to take a skills-based approach. The intervention was designed with community input following the SWIFT trial and uses the extended parallel process model to build on the teachable moment of stroke diagnosis.⁴⁸ Using the extended parallel process model, intervention participants’ perceived risk and perceived efficacy are evaluated to build skills for sustained behavior change. DESERVE focused on developing strategies and tools that empower patients to promote active communication with their physicians, reduce barriers to medical adherence, and enhance vascular health. For example, we helped participants identify areas of confusion and create a checklist of questions for their physicians, and we reinforced these skills prior to physician visits through telephone calls. Optimizing skills in this way may be more effective in achieving control of risk factors.⁴⁹ By training patients to take ownership of controlling risk factors, DESERVE allows the process to be sustainable beyond the health care system. Additionally, unlike most secondary prevention interventions to date and to our knowledge, our materials were health literate and created in partnership with community members. Health literacy has particular relevance for individuals with chronic illness because they need to manage their disease over an indefinite period of time.

Strengths and Limitations

This study had several strengths. First, by using an RCT design, we were able to rigorously evaluate the efficacy of our intervention. Second, through using bilingual coordinators and working with community partners, we were able to enroll and engage patients with stroke/TIA who are often underrepresented in RCTs.⁵⁰ Our diverse study sample also allows us to generalize our results to other urban settings. Additionally, the study benefited from a high retention rate, which minimized the potential for selection bias, which can be a particular threat to prospective studies.

This study also had several important limitations. First, as part of usual care in our trial, we ensured that every participant received health literate and linguistically appropriate educational materials, which may not be a consistent practice in real-world settings. This enhancement may have attenuated the effect of the intervention. Second, owing to the difficulty of scheduling appointments, we were not able to obtain in-person follow-up BP measurements for some of our participants and had to rely instead on physician records. These measurements may not have been taken using the same standardized instruments and procedures as indicated in our study protocol. However, the research assistant collecting measurements from physicians was blinded to intervention status, so any misclassification of the outcome would be nondifferential leading to underestimated results. Additionally, the distribution of source of follow-up BP measurement did not differ significantly between trial arms in the overall sample or between trial arms by race/ethnicity, reducing potential for bias.

Conclusions

While the overall trial demonstrated SBP reductions in both arms leading to a nonsignificant intervention effect, DESERVE results suggest that culturally tailored, skills-based education with follow-up reinforcement may lead to clinically meaningful BP reduction among certain racial/ethnic populations 1 year poststroke. This is important as few behavioral intervention studies in individuals who have had stroke have reported significant long-term differences in vascular risk reduction, and fewer have focused on a skills-based approach. Further, DESERVE is one of the first interventions potentially demonstrating capacity to decrease racial/ethnic disparities in stroke risk reduction. Our results suggest that culturally tailored, skills-based interventions may be more useful than knowledge-focused interventions in achieving sustained vascular risk reduction and addressing racial/ethnic disparities in stroke; however, these findings should be tested in future studies.

Supplement.

Trial protocol.