Technical Dissonance in Home Blood Pressure Monitoring After Stroke: Having the Machine, but Not Using Correctly

Abstract

Background

In individuals with hypertension (HTN), lowering blood pressure (BP) after a stroke can lower the risk of stroke recurrence, but many patients do not reach the goal. Home blood pressure monitoring (HBPM) can help patients get to the goal, but rates of use and quality of technique have not been evaluated.

Methods

We conducted a cross-sectional study of patients with stroke. Patients were eligible if they had a stroke within 2 years, had HTN, and lived at home. We classified patients as correctly performing HBPM if they used an arm cuff, sat ≥ 1 min before measurement, took ≥ 2 measurements, and use within 6 months. The primary outcome was the proportion of patients who had an HBPM and used it correctly, which we calculated according to race and ethnicity. We also asked patients what they would do if they found results outside the goal.

Results

Among 150 participants, 120 (81%) possessed an HBPM and 29 (21%) used it correctly. We observed no significant disparity in rates of possession or correct use between non-Hispanic White participants and participants from underrepresented groups. Seventy percent of non-Hispanic White patients said they would contact their provider if their BP was above goal vs. 52% of underrepresented patients (P = 0.21).

Conclusions

Most patients after stroke have an HBPM, but only about 1 in 5 use it correctly. Approximately half of the patients from underrepresented racial or ethnic groups do not have a plan for responding to the values above goal. Our results indicate opportunities to improve the dissemination and correct use of HBPM.

Graphical Abstract

Graphical Abstract.

Hypertension (HTN) affects approximately 70% of patients with a recent stroke and treatment is highly effective in reducing the risk of stroke recurrence.¹ In the PROGRESS trial of 6,000 hypertensive individuals with ischemic or hemorrhagic stroke from Asia, Australasia, and Europe, treatment based on a combination of perindopril and indapamide reduced risk for fatal or nonfatal stroke by 28%.² In another study, investigators calculated that every 10-mmHg reduction in systolic blood pressure (BP) was associated with a 33% reduction in risk of recurrent stroke.³ Therefore, BP lowering is effective for secondary stroke prevention.

Currently, the American Heart Association (AHA) recommends an office BP goal of <130/80 mmHg for most patients with HTN who experience a stroke.¹ Despite this recommendation, many individuals remain above goal following an ischemic or hemorrhagic stroke.⁴ By several estimates, at least 55% of patients are above 140/90 mm Hg at stroke clinic follow-up visits.^5,6 Differences in BP control also contribute to existing racial and ethnic disparities among patients with stroke. Only 39.3% of Black patients with a history of stroke have controlled BP compared to 50.3% of White patients.⁷ Improved diagnosis and management of HTN, therefore, represents an opportunity to reduce risk of recurrent stroke and address health disparities.

One strategy for improving BP control among patients with stroke is home blood pressure monitoring (HBPM). HBPM is simple and inexpensive; almost every patient (or caregiver) can be taught proper technique. Modern machines store results and many can connect wirelessly to computers or the internet for reporting to healthcare professionals. Compared with usual care that involves only office-based BP monitoring, HBPM is associated with significant reductions in systolic and diastolic BP as well as a reduction in therapeutic inertia.⁸ A recent feasibility study found that among 50 patients with stroke from under-resourced communities, BP control was better among patients who received nurse-supported home BP telemonitoring compared with usual care.⁹ Notably, HBPM has been associated with a reduced risk of stroke recurrence in observational studies.¹⁰

In 2017, the Writing Group for the AHA HTN guideline included a class 1A recommendation for use of HBPM in all capable patients with HTN.¹¹ The American Diabetes Association also recommends HBPM for all patients with diabetes and HTN. The European Society of Hypertension also recommends HBPM to confirm the diagnosis of HTN and monitor patients.^12,13

Given poor control of BP after stroke and existing recommendations for HBPM for patients with hypertension, it may be useful for healthcare professionals to know the extent and quality of HBPM use among patients with prior stroke. The aims of this study, therefore, were to (i) estimate the proportion of patients with a recent stroke who have an HBPM and use it correctly, (ii) examine racial and ethnic disparities in HBPM use, and (iii) identify barriers to HBPM use.

METHODS

This was a cross-sectional study. Eligible patients were admitted to Yale New Haven Hospital (YNHH) with an acute (ischemic or hemorrhagic) stroke within two years of their interview. They also had a history of HTN, lived at home, and were able and willing to provide informed verbal consent. We excluded patients with dementia given the need to participate in a detailed interview and we excluded patients who did not speak English or Spanish because we did not have access to interpreters for other languages. A patient was classified as having HTN if this diagnosis was listed in their medical record. They were not required to be taking antihypertensive medications. We excluded patients who resided in a rehabilitation or other long-term care facility because this may affect BP monitoring (Supplementary Table S1online).

At the beginning of the research, participants were recruited from a longitudinal outcome study maintained at YNHH (York Street campus) for patients admitted to the inpatient stroke service. Staff invited eligible patients to join our study during follow-up calls at 3 and 6 months after discharge. However, when it became apparent that we would not meet our enrollment goal in the 9-month recruitment phase, recruitment was expanded to patients who were contacted post-hospitalization by Yale stroke nurse navigators and patients who attended the Yale stroke clinic.

Data sources included the electronic medical record (for date of birth, sex, and medical history) and a structured patient interview. Interviews took place between 22 July 2021 and 4 April 2022. In addition to questions about HBPM, the patient interview (see Supplementary Material online) gathered participant-reported information on race, ethnicity, medical insurance, work status, and included standard screens for depression, social isolation, loneliness, and HTN self-efficacy (i.e., a person’s belief in their capacity to complete behaviors needed to reach specific goals).^14–18

The primary outcome was the proportion of patients who reported that they had an HBPM and had used it in correctly at least once within the preceding 6 months. Patients were classified as having an HBPM if they reported any device at home which they could access for their own use. We classified a patient as using their HBPM correctly if they used an arm cuff (rather than a wrist or forearm cuff), sat ≥ 1 min before first measurement, took ≥ 2 measurements per sitting, and used within 6 months. We modified these criteria from the AHA guidelines for HBPM.¹⁹ In contrast to those guidelines, we required sitting ≥ 1 min rather than ≥5 min before first measurement, we did not require abstinence from caffeine, nor did we require positioning the cuff at the level of the heart. We used these less stringent criteria because, based on clinical experience, we suspected that participants would have difficulty reporting some of the AHA criteria accurately (e.g., last caffeine use and relation of the cuff to the heart) and that few patients would meet all the criteria. We were confident, however, that a finding of adherence to more minimal criteria for correct use would be meaningful.

Our sample size was determined to meet 2 of our research goals. First, assuming the rate of correct HBPM use was 20%, estimating this proportion with 5% absolute precision and 80% confidence required enrolling 106 patients. The estimate of 20% correct use was derived from a casual survey of neurologists and general internists at Yale University.

Second, to compare the use of HBPM by racial and ethnic groups, we increased the sample size and oversampled certain race/ethnic groups. We hypothesized that 30% of White/non-Hispanic patients would use an HBPM correctly and 10% of non-White or Hispanic patients (“underrepresented”) would use an HBPM correctly. To have 80% power to detect a difference of this magnitude at a 5% significance level, we needed 62 patients in each group (124 total). We increased the final prespecified sample to 150 total (75 per group) to account for errors in the assumptions underlying our sample size calculation. The oversampling of underrepresented patients was accomplished by only approaching underrepresented individuals (as listed in the demographics section on their electronic medical records) after 75 non-Hispanic White patients had been enrolled. Underrepresented individuals were those who identified their race as non-White (i.e., American Indian or Alaska Native, Asian, Black or African American, Native Hawaiian, or other Pacific Islander) or their ethnicity as Latino or Hispanic per the NIH (National Institute on Minority Health and Health Disparities) definition for minority health groups.²⁰

Baseline features were described using means, medians, and percentages as appropriate. Chi-square tests were used to compare the group differences in proportions.

RESULTS

Based on the dates of our first and last interview, patients were eligible for the study if they were discharged between 22 July 2019 and 4 April 2022 after an ischemic or hemorrhagic stroke. During this window of eligibility, 2,873 patients with ischemic or hemorrhagic stroke were discharged from YNHH (69% White, 18% Black, 2% Asian and Pacific Islander, 3% Other race, and 8% Hispanic or Latino). Using a convenience sample approach, 176 of these patients who met the inclusion criteria were invited to participate and 150 patients were accepted (Figure 1). Female patients were more likely to decline participation, as were patients with a history of prior stroke, although rates of comorbid conditions (except for diabetes) were higher among participants compared to those who refused (Supplementary Table S2 online). Four participants who declined to disclose race and ethnicity were excluded from analyses dependent on these factors.

Figure 1.

| Cohort formation.

The mean age in the non-Hispanic White group was 71.5 years (SD 10.5) compared with 60.7 years (SD 13.3) in the underrepresented group. Both groups had a male predominance (66% in the non-Hispanic White group and 54% in the underrepresented group). Of the underrepresented participants, 48 (67%) identified as Black and 17 (24%) participants identified as Hispanic (Table 1).

Table 1.

| Characteristics of participants, overall and by race/ethnicity group

Characteristic	All (n = 150)	White, N-H (n = 74)	UR (n = 72)
Age, years, mean (SD)	66 (13)	71.5 (10.5)	60.7 (13.3)
Male, n (%)	90 (60%)	49 (66%)	39 (54%)
Race, n (%)
White	82 (55%)	74 (100%)	8 (11%)
Black	48 (32%)		48 (67)
Asian	5 (3%)		5 (7%)
Native Hawaiian/other Pacific Islander	1 (1%)		1 (1%)
More than 1 race	3 (2%)		3 (4%)
Refused/uncertain	11 (7%)		7 (10%)
Hispanic ethnicity, n (%)	17 (11%)		17 (24%)
Takes BP medications	140 (94%)	70 (95%)	66 (92%)
Medical history, n (%)
Prior ischemic stroke	26 (17%)	6 (8%)	20 (28%)
Prior hemorrhagic stroke	2 (1%)	1 (1%)	1 (1%)
Diabetes	59 (39%)	25 (34%)	33 (46%)
Coronary Artery Disease (CAD)	36 (24%)	16 (22%)	20 (28%)
Congestive Heart Failure (CHF)	25 (17%)	7 (9%)	18 (25%)
Chronic Kidney Disease (CKD)	30 (20%)	10 (14%)	20 (28%)
Stroke type, n (%)
Ischemic	137 (91%)	73 (99%)	61 (85%)
Hemorrhagic	13 (9%)	1 (1%)	11 (15%)
Months since the stroke, median (range)	4.4 (0.1–30.2)	4.8 (0.1–18.2)	4.0 (0.1–30.2)
Psycho/social
Major depressive disorder (Patient Health Questionnaire (PHQ)-2: 3+)	28 (19%)	12 (17%)	15 (21%)
Poor HTN self-efficacy (score <9)	63 (53%)	31 (53%)	32 (56%)
Socially isolated (score 2+)	72 (50%)	31 (43%)	40 (58%)
Health insurance, n (%)
Medicaid	33 (22%)	5 (7%)	28 (39%)
Medicare	62 (41%)	39 (53%)	20 (28%)
Commercial	53 (35%)	30 (41%)	22 (31%)
None	2 (1%)	0 (0%)	2 (3%)
Work status, n (%)
Working full time	25 (17%)	11 (15%)	13 (18%)
Working part-time	7 (5%)	6 (8%)	1 (1%)
Not working	19 (13%)	3 (4%)	16 (22%)
Retired	71 (47%)	48 (65%)	21 (29%)
On disability	27 (18%)	6 (8%)	21 (29%)

Abbreviations: BP, blood pressure; HTN, hypertension; N-H, Non-Hispanic; UR, underrepresented.

Missing data (all, White/N-H, UR): Hispanic ethnicity (1, 0, 0); prior ischemic stroke (1, 0, 1); CKD (1, 1, 0); major depressive disorder (5, 2, 2); Self-efficacy (32, 16, 15); social isolation (5, 2, 3); work status (1, 0, 0).

A higher percentage of non-Hispanic White participants had an ischemic stroke compared with participants from underrepresented groups (99% vs. 85%). Underrepresented participants had higher rates of almost all medical comorbid conditions, including prior ischemic stroke (28% vs. 8%), diabetes (46% vs. 34%), coronary artery disease (28% vs 22%), congestive heart failure (25% vs 9%), and chronic kidney disease (28% vs. 14%).

More non-Hispanic White participants were retired compared to participants from underrepresented groups (65% vs. 29%), consistent with observed age differences between the groups. In the underrepresented group, 51% of participants were disabled or not working compared to 12% from the non-Hispanic White group. Additionally, 39% of participants had Medicaid insurance from the underrepresented group compared to 7% in the non-Hispanic White group. Rates of major depressive disorder and poor BP self-efficacy were similar between groups. Finally, 56% of underrepresented participants were socially isolated, compared to 42% of the non-Hispanic White group.

Among all participants, 81% (95% CI, 73% to 87%) possessed an HBPM with no significant difference between groups (84% non-Hispanic White vs. 76% underrepresented; P = 0.28) (Table 2). Among all 150 participants, 21% (95% CI, 14% to 27%) had an HBPM and used it correctly with no significant difference between groups (21% non-Hispanic White vs. 18% underrepresented; P = 0.60).

Table 2.

| HBPM use, overall and by race/ethnicity group

Feature, n (%)	All (n = 150)	White, N-H (n = 74)	UR (n = 72)	P
Possesses a HBPM	120 (81%)	61 (84%)	55 (76)	0.28
Uses a HBPM (any)	109 (77%)	57 (81%)	48 (72%)	0.18
Uses HBPM correctly	29 (21%)	15 (21%)	12 (18%)	0.60
Established BP goal without provider	54 (38%)	20 (28%)	32 (47%)	0.02
If yes and have HBPM, Would contact the provider if the above goala	30 (63%)	14 (70%)	14 (52%)	0.21

Abbreviations: HBPM, home blood pressure monitoring; N-H, non-Hispanic; UR, underrepresented.

Missing data (all, White/N-H, UR): Possesses a HBPM (1, 1, 0); Uses a HBPM (9, 4, 5); Uses HBPM correctly (9, 4, 5); Established BP goal (7, 3, 4).

^aDenominator (all, White/N-H, UR): 48, 20, 27 participants.

When examining the individual criteria for correct use, 84% of participants used an arm cuff, 50% waited >1 min prior to taking their BP measurement, and 57% took more than one BP measurement per sitting. When comparing groups, we found no statistically significant difference in anatomic placement or wait time for BP measurement between groups. The proportion of participants taking at least 2 measurements was 39% of non-Hispanic White participants compared to 54% of the underrepresented participants (P = 0.11) (Table 3).

Table 3.

| Criteria for correct HBPM use, overall and by race/ethnicity group

Criterion, n (%)	All (n = 109)	White, N-H (n = 57)	UR (n = 48)	P
Anatomic placement
Forearm or wrist	17 (16%)	10 (18%)	7 (15%)	0.68
Arm	92 (84%)	47 (82%)	41 (85%)
Wait time
None or <1 min	54 (50%)	29 (51%)	23 (48%)
1–5 min	55 (50%)	28 (49%)	25 (52%)	0.76
No. of measurements
1	57 (52%)	35 (61%)	22 (46%)
2+	52 (48%)	22 (39%)	26 (54%)	0.11

Abbreviations: HBPM, home blood pressure monitoring; N-H, Non-Hispanic; UR, underrepresented.

The following participant features were associated with numerically lower rates of correct HBPM use (none statistically significant): unemployed status, having Medicaid insurance, having major depressive disorder, poor BP self-efficacy, and being socially isolated (Table 4).

Table 4.

| Correct HBPM use by participant features

Feature	% Correct HBPM use
	Feature present	Feature absent	P
Not employed	15%	17%	0.89
Medicaid insurance	9%	17%	0.25
Major depressive disorder	15%	22%	0.47
Poor self-efficacy	21%	27%	0.42
Socially isolated	18%	25%	0.34

Abbreviation: HBPM, home blood pressure monitoring.

Twenty-eight percent of non-Hispanic White participants had established a BP goal with their provider compared with 47% of participants from underrepresented groups (P = 0.02). Seventy percent of non-Hispanic White participants who stated that they had a goal said they would contact their provider if their BP was above that goal compared with 52% of underrepresented participants (P = 0.21) (Table 2).

Among 150 patients in our sample, 30 (20%) did not have a BP monitor at home. (Table 5). Compared to patients who did have a machine at home, patients without one were significantly more likely to report that cost was a barrier (P = 0.04) and that they had not been advised by a healthcare professional to check their BP at home (P = 0.0002; for comparisons between participants with and without a machine).

Table 5.

| Barriers to HBPM for participants with and without HBPM machine

Barrier, n (%)	With machine (n = 120)	Without machine (n = 29)	P
Cost of machine	7 (6%)	5 (18%)	0.04
Time for HBPM	8 (7%)	1 (4%)	0.53
Not advised to do	6 (5%)	8 (28%)	0.0002
Fear or anxiety	10 (9%)	1 (4%)	0.37
Pain or discomfort	1 (1%)	0%	0.62
Fit of cuff	4 (3%)	0%	0.32
Other	16 (14%)	7 (24%)	0.16
Any of above	39 (33%)	19 (66%)	0.001

Abbreviation: HBPM, home blood pressure monitoring.

Missing data (with machine, without machine): cost (2, 1); time (2, 1); not advised (2, 0); fear or anxiety (3, 1); pain or discomfort (3, 1); fit of cuff (2, 1); other (2, 0).

DISCUSSION

Among 150 patients with a recent stroke in our convenience sample, 81% possessed an HBPM and 77% used it at least once within the past 6 months; however, only 21% used it correctly according to minimal criteria modified from recommendations by the AHA. We observed no significant differences in rates of possession or correct use between non-Hispanic White participants and participants from underrepresented groups. In addition, we found no significant association between correct use and employment status, insurance type, depression, self-efficacy, or social isolation. This is the first study to the best of our knowledge that evaluates the degree of correct use of HBPM among patients with stroke.

A total of 30 study participants reported that they did not possess an HBPM. Compared with participants who did have a monitor, these participants were significantly more likely to report that cost was a barrier and that they had not been advised to use one. These 2 findings may represent an opportunity to improve the use of HBPM by increasing access to a device and educating patients directly or encouraging healthcare professionals to implement professional guidelines that recommend HBPM for all patients with HTN. Importantly, our findings for incomplete use of HBPM should be interpreted in the context of policies of our institution, which do not include provision of an HBPM on discharge or standardized instructions for their use.

More participants from underrepresented groups had established a BP goal with their healthcare professionals (47% compared to 28% of non-Hispanic White individuals, P = 0.02); however, more non-Hispanic White participants reported that they would contact their provider if the above goal compared with participants from underrepresented groups (70% compared to 52%, P = 0.21). We did not have data to explain these differences.

There are several strengths of this study. The study was conducted according to a written protocol and structured interviews conducted by 2 trained researchers who coordinated efforts to minimize variation in data collection. Our sampling strategy led to the enrollment of a population enriched from historically disenfranchised groups.

There are also some limitations. We set out to enroll consecutive patients called for follow-up interviews in a YNHH longitudinal outcome study. This strategy was designed to reduce the risk of selection bias; however, we converted to a convenience sample after slow enrollment. Another limitation is that we included patients who were immediately discharged home following their stroke event; 25% of participants were interviewed within 2 months after the event. This potentially may not have allowed adequate time to obtain or be advised to use HBPM. We relied on volunteers for Spanish interpretation; therefore, we could not enroll Spanish-speaking patients when these interpreters were not available. This led to a lower number of Hispanic individuals in the cohort compared to participants from other underrepresented groups. There are multiple classification systems for underrepresented race and ethnicity status; we opted to use the NIH definition. Had we used a different system, our results might have been affected but we believe any affect would be minimal. Finally, our criteria for correct use of HBPM were less stringent than those recommended by most professional guidelines; our estimate of 21%, therefore, might be interpreted as a “best case” scenario.

The findings of this study indicate widespread incorrect use of HBPM among patients with prior strokes. Next steps in building upon this finding include creating a larger study with a sampling strategy that reduces risk of bias and error and allows for identification of personal characteristics associated with use and nonuse. Candidate variables that we believe may be associated with use include having Medicaid insurance, major depressive disorder, poor BP self-efficacy, and social isolation. In our research, these factors were more prevalent among patients who did not engage in HBPM compared with patients who did engage in HBPM, but our sample size was not large enough to demonstrate statistical significance in the differences.

If our findings for low rates of correct use of HBPM among patients with stroke are confirmed, it will be important to design and test interventions that improve rates of correct use. Potential interventions include hospital-based or clinic-based education for patients with stroke and their caregivers to improve rates of correct use of HBPM. Similar initiatives are in place for other chronic diseases, such as diabetes. For instance, diabetes self-management education, which is taught by certified diabetes educators, is effective for preventing diabetes complications, enhancing health outcomes, and improving knowledge and self-efficacy.²¹ Coupling education with goal-setting, treatment algorithms, and team-based care has improved BP control in patients without stroke and could be effective in those with stroke. These efforts could be augmented by the use of machines that send data directly to healthcare providers for timely review and titration of BP medications.

Finally, our findings on barriers to HBPM use suggest that providing patients with free access to monitors may help those who cannot afford them. Pending legislature in certain states in the United States, e.g., would provide monitors to Medicaid recipients with uncontrolled BP.²²

SUPPLEMENTARY MATERIAL

Supplementary materials are available at American Journal of Hypertension (http://ajh.oxfordjournals.org).

DISCLOSURE

Dr de Havenon has received investigator-initiated clinical research funding from the NIH (K23NS105924), AAN, Regeneron, AMGEN, and AMAG pharmaceuticals, has received consultant fees from Integra and Novo Nordisk, royalty fees from UpToDate, and has equity in TitinKM and Certus. Dr Sansing reports funding from NINDS and the AHA. Dr Sharma reports funding from NINDS (K23NS121634).