Mobile Health Intervention to Close the Guidelines-To-Practice Gap in Hypertension Treatment: Protocol for the mGlide Randomized Controlled Trial

Background Suboptimal treatment of hypertension remains a widespread problem, particularly among minorities and socioeconomically disadvantaged groups. We present a health system–based intervention with diverse patient populations using readily available smartphone technology. This intervention is designed to empower patients and create partnerships between patients and their provider team to promote hypertension control. Objective The mGlide randomized controlled trial is a National Institutes of Health–funded study, evaluating whether a mobile health (mHealth)-based intervention that is an active partnership between interprofessional health care teams and patients results in better hypertension control rates than a state-of-clinical care comparison. Methods We are recruiting 450 participants including stroke survivors and primary care patients with elevated cardiovascular disease risk from diverse health systems. These systems include an acute stroke service (n=100), an academic medical center (n=150), and community medical centers including Federally Qualified Health Centers serving low-income and minority (Latino, Hmong, African American, Somali) patients (n=200). The primary aim tests the clinical effectiveness of the 6-month mHealth intervention versus standard of care. Secondary aims evaluate sustained hypertension control rates at 12 months; describe provider experiences of system usability and satisfaction; examine patient experiences, including medication adherence and medication use self-efficacy, self-rated health and quality of life, and adverse event rates; and complete a cost-effectiveness analysis. Results To date, we have randomized 107 participants (54 intervention, 53 control). Conclusions This study will provide evidence for whether a readily available mHealth care model is better than state-of-clinical care for bridging the guideline-to-practice gap in hypertension treatment in health systems serving diverse patient populations. Trial Registration Clinicaltrials.gov NCT03612271; https://clinicaltrials.gov/ct2/show/NCT03612271 International Registered Report Identifier (IRRID) DERR1-10.2196/25424

1 R01 HL138332-01A1 3 ZRG1 HDM-G (59) LAKSHMINARAYAN, K months and 12 months after randomization. Aim 2 will examine mGlide usability for providers and provider experience and satisfaction with mGlide. It will also examine whether medications are managed differently for participants in the two groups. Aim 3 will examine whether patients are more satisfied with care in the mGlide group, whether they are more "activated" and have a greater sense of self-efficacy in managing their HTN. Aim 4 will be a cost-effectiveness analysis of providing mGlide care. Our long-term goal is to prevent stroke and improve cardiovascular health in populations by increasing health system efficiency and effectiveness. The results from our mGlide RCT will provide evidence for the use of readily available mHealth technology for bridging the guideline-to-practice gap in HTN treatment in low resource health systems serving minority groups.

PUBLIC HEALTH RELEVANCE:
This study addresses the significant public health problem of uncontrolled hypertension and can prevent stroke and improve cardiovascular health while reducing costs. The current U.S. annual estimated cost of hypertension is $51 billion. Results from this study will provide evidence for the use of readily available mobile health technology for bridging the guideline-topractice gap in hypertension treatment. Our study will be implemented in low resource health systems serving minority groups. We also include an economic analysis showing the cost-effectiveness of our mobile health technology based care model.

CRITIQUE 1
Significance: 3 Investigator(s): 1 Innovation: 2 Approach: 2 Environment: 1 Overall Impact: Control of hypertension is important for a number of reasons but remains suboptimal. This is a 12-month RCT (6-month active intervention) using mHealth wireless Self Monitored BP measurements fed into a semi-automated decision support tool (called mGlide) compared to state of clinical care targeting 3 minority groups with low rates of hypertension control. The primary outcome is "rate of HTN control". The work is grounded in the premise that HTN control can be improved if one can affect two changes: 1) get people to self-monitor BP (SMBP) and 2) get clinicians to closely and actively adjust medications. While the causes of the "guideline to practice" gap in HTN control are undoubtedly complex their premise is supported by high quality evidence. The approach used has high scientific rigor and use of wireless monitoring is innovative as is use of mGlide. A successful outcome (positive result) would be highly significant in that an approach to improved HTN control in one or several of their high-risk minority groups would provide generalizable guidance for others to follow in similar populations. Sex as a bio variable was adequately addressed. The resubmission addresses the major concerns of the first review. Overall the study is considered likely to have high impact.

Strengths
 The premise is supported by prior work cited in the literature.  Their preliminary results from a 50-person short term study indicate that it is likely that patients will use the monitor and report data (at least for a while), a key point as digital home monitors have been around for some time but patient use is often irregular. Preliminary data even suggest efficacy although the study was not long term. The underlying premise is that if one can get patients to reliably stream the measurements an activation and a degree of accountability are created---and under such conditions it is likely patient compliance with meds will be good.  Successful demonstration of improved control with this approach would be very significant as smart phones are very prevalent and the decision support approach can be scaled and translated to health systems of varying sizes and configurations.

Weaknesses
 Minor: data supporting the idea that the flow of BP information will be both regular and sustained are limited (89% of 120 days with no information on trend).  Minor: Data on physician acceptance and smoothness of workflow integration are still being collected.  Minor. There is no justification for the assumption that the mHealth and wireless monitor will be used regularly when portable digital monitors do not seem to be. Also, it is less clear how important physician management (one of the outcomes is how often changes in meds are made---Is it the meds or the fact that patients don't take the meds?).

Strengths
 The full complement of investigators needed to carry out this work are in place including: focused PI who, despite being a New Investigator, has completed the preparatory pathway (K23) toward independence in this sort of research and has experienced guidance (e.g. Russell Luepker and many others); experienced clinical trials biostatistician (John Connett); clinical community (including minority experience) based co investigators/trialists (Kathleen Culhane-Pera and Shannon Pergament); cost effectiveness analysis (John Nyman); Pharmacy (Sarah Marie Westberg); large trials involving hypertension (Russell Leupker) and others.
Weaknesses  None noted.

Innovation:
Strengths  mHealth and wireless BP monitor.  mGlide decision support pathway.

Weaknesses
 Minor: The lack of significant participation of the health care system which seems to play the role of allowing the research to go forward rather than finding novel ways to support (see Approach) it is not novel.

Approach:
Strengths  The individualized RCT approach is seen as a strength.
 The evaluation of intervention impact on providers, patients and costs is seen as a strength.
 The comparison arm is well described.
1 R01 HL138332-01A1 5 ZRG1 HDM-G (59) LAKSHMINARAYAN, K  Approach is seen as properly navigating the tension between having a "multilevel" approach and keeping the intervention focused and finite enough so that conclusions can be drawn from the results.

Weaknesses
 Minor: The design lacks incentives for the physicians.

Environment:
Strengths  This is a very strong environment with great experience in epidemiology, public health and clinical trials---and a community based practice network with appropriate access to and experience with the target minority groups.

Weaknesses
 The previously noted deficiencies in the Protection of Human Subjects and DSMB configuration have been addressed and the age inclusion range widened. Overall Impact: This proposal focuses on automated home blood pressure monitoring to improve hypertension control and reduce the risk of stroke. Hypertension and increased risk of stroke are clearly very common and important health concerns and fit well within FOA's specified subjects for consideration. Both the PI and Co PIs are well trained and published in this area. A pilot study apparently has been completed, but is inadequately presented in this application. Data important to the success of the study is often cited as references to the literature or to even biosketches, instead of a one or two sentence summary to support the concept being discussed. This raises concerns whether there is adequate focus and organizational capacity to effectively implement a complex study regarding the use of automated, home blood pressure monitoring to achieve normal blood pressures in hypertensive patients. This reviewer also has significant concerns about recruitment of patients from the Fairview Health System, which is not well described in the protocol. The Fairview web page describes six hospitals, 39 primary care clinics, 34 pharmacy locations, and 4000 doctors and providers in 30 cities across Minnesota. If typical of university affiliated health care systems, it is unlikely that Fairview has a focus on health disparities or economically distressed patients, as opposed to the two FQHCs included in the protocol. This reviewer was unable to find criteria in the protocol for patient recruitment that focuses on health disparities within the Fairview system. Data from table 8 show that 93% (22,033/23,687) of the "high risk" (not defined in the protocol) patients with uncontrolled hypertension are available in the Fairview system, just 7% in the FQHCs; this very large difference is not credible. Although recruitment will be limited to 11 Fairview clinics, there is no discussion of focusing on health disparities. Both the control and intervention arms will do home BP monitoringalthough using different monitoring systems with electronic transmission back to the research team occurring only in the intervention arm, which includes a pharmacist. Engagement of the primary care provider appears to be similar in the two arms. It appears that the electronic feedback of BPs does not get to the electronic health record and available to non-study care providers, except as recorded in text notes by the study pharmacist.

Protections for Human Subjects:
.

Significance:
Strengths  Hypertension is a common health problem, and better control will reduce the risk stroke. 1 R01 HL138332-01A1 7 ZRG1 HDM-G (59) LAKSHMINARAYAN, K Weaknesses  Page 159, Specific Aim 1, MODERATE: The statement, "rates of HTN treated into optimal range in the 2 groups at 6 months and 1 year after randomization", does not specify which of the many blood pressure measurements available to the study team will be used to assess this comparison.  Specific Aim 2 is not very specific, MODERATE: a) What is meant by provider actions on anti-HTN medications --new prescriptions or prescription dose changes? Why are focus groups regarding provider experience included on the same aim?
 Same for Specific Aim 3, MODERATE; At least a dozen comparisons are listed.
 Specific Aim 4, MAJOR: Why are cost-effective analyses (CEA) being done using data from the literature?
 MODERATE, Specific aim 4 states, "CEA will model the experience of the two trial arms over a 10-years" however, Specific Aim 1 says that this is a "12-month, 2-arm RCT".  MAJOR, The Impact statement at the bottom of page 159 states, "This study can … prevent stroke …." This is an incorrect statement; stroke incidence is not being measured.

Investigator(s):
Strengths  The investigative team has published extensively in the area of hypertension control.
Weaknesses  None noted.

Innovation:
Strengths    Page 162 MODERATE: The difference between "mhealth facilitated self-monitoring" in the intervention group and "digital monitoring" in the control group is not clear.  MAJOR: Table 5 presents data that two (WEST SIDE COMMUNITY HEALTH SERVICE and Community University Health Care Center) of the three recruitment health sites focus their care on disadvantaged patients (i.e. are FQHCs), but not Fairview Health Services, which appears to be a consortium of hospitals and clinics owned by or closely affiliated with the University of Minnesota. We are not told how patient selection from Fairview will focus on disadvantaged patients.

Weaknesses
 Page 166, 6.5.3: Eligible stroke survivors, MODERATE: Referencing a clinical practice guideline for incidences of hypertension and discharge home among stroke survivors is weak; it would have been much better if these data came directly from the population being considered for study. Also, it would be better if all of these study participants were recruited from the FQHCs, where patients are more likely to meet the definition of having a health disparity, rather than the Fairview Health System.  Page 166, 6.5.4 Eligible high risk primary care participants: MAJOR: The term, high risk, is used throughout this proposal without defining how the patient values will be obtained. High risk is defined as "> 7.5% CVD risk / 10 years" citing stroke statistics from the American Heart Association (reference 56). However, the data elements required for this risk estimate are not provided; nor are we told how these data will be acquired.  Page 166, Table 8, MODERATE: The inclusion of 66,768 as "eligible primary care patients at high risk" from the Fairview Health Care system is misleading. We are not given data on the characteristics of the 48 Fairview primary clinics, but one has to presume that they are not FQHCs and represent a heterogeneous group of private care ambulatory facilities caring for patients not insured by Medicaid or otherwise economically disadvantaged.
 Page 166, 6.7.1 Intervention description, MAJOR: Details on remote monitoring of BP are very sketchy; e. g. What is the frequency of the BP measurement? What kind of cuffs are used? How are spurious values dealt with? It would have been much better to include the PI's description here of the Withings Remote Monitoring System rather than refer to a letter from company that makes the device.  Page 167, Table 10 Comments, NINDS CREST2, MODERATE: The cited reference, (52), a web page for the CREST 2 trial, does not describe a protocol for measuring BP. It would have been better to summarize a protocol for measuring BP here: e.g., body position during measurement, sitting quietly with no significant exercise for several minutes, avoidance of isometric muscle contraction, etc. The American Heart Association web page, Monitoring Your Blood Pressure at Home (http://www.heart.org/HEARTORG/Conditions/HighBloodPressure/KnowYourNumbers/Monitorin g-Your-Blood-Pressure-at-Home) is an excellent source for this information.
 Page 167, 6.7.3 Participant contact, MODERATE: The monthly (first six months) and q 2 month contact with participants in both study arms is not defined. Will the contact be the same for both the intervention and control arms?
 Page 167-8 Table 10 and Table 10 Comments, MODERATE: These are very difficult to comprehend; for example, consecutive sentences in the Comments deal with two very disparate concepts: Varimax rotation in principal factor analysis and focus groups on barriers and solutions for translation of mGlide into clinical practice.
 MODERATE: Table 10 and Table 10 comments encompass nearly a full page of the seven pages devoted to 6.0 RESEARCH DESIGN AND METHODS.
 Page 168, 6.9 Randomization and blinding. MODERATE: The protocol states that randomization will be "site-specific", but "site: is not defined. This is important because the number of randomization blocks is dependent on this definition.  6.10 Data coordination, management, missing values, quality control MAJOR: This critically important facet of the study is incomplete with only nine lines of text.