Hypertension Improvement Project (HIP): study protocol and implementation challenges

Approximately 50 million adult Americans have hypertension [1], which is the most prevalent risk factor for cardiovascular and kidney disease [2] and accounts for approximately 35% of atherosclerotic cardiovascular disease (CVD) [3]. Fortunately, the treatment of hypertension reduces associated risks: lowering blood pressure (BP) by a variety of strategies reduces the risk of stroke by approximately 35%, congestive heart failure by 42%, and coronary heart disease by 28% [4‐7]. Therefore, an important key to reducing the burden of hypertension-related CVD is to increase the proportion of patients who achieve optimal BP control. To promote this goal, the Joint National Committee on Prevention, Evaluation, Detection, and Treatment of High Blood Pressure (JNC) has established guidelines that have been widely disseminated to practitioners through the National Institutes of Health High Blood Pressure Education Program and other organizations [8].

The latest JNC guidelines (JNC-7), published in 2003, classify BP into stages and provide recommendations for treatment and follow-up. The recommended goal for patients with uncomplicated hypertension is an average systolic BP of less than 140 mm Hg and a diastolic BP of less than 90 mm Hg. For patients with certain comorbid conditions that increase risk (i.e., diabetes mellitus and chronic kidney disease), the goal BP is less than 130/80 mm Hg [8].

Treatments for high blood pressure are classified as lifestyle modification or medication. The guidelines direct physicians to prescribe lifestyle modification to all patients with BP that is above the optimal category (> 120/80 mm Hg). The recommendations for lifestyle modification are evidence-based and include adoption of a healthy dietary pattern such as the Dietary Approaches to Stop Hypertension (DASH) diet [9], losing weight if overweight, reducing sodium intake, increasing physical activity, and limiting alcohol intake [8]. Each of these behavioral interventions has been shown to lower BP and contribute to hypertension control. For medication therapy, the guidelines indicate when and whom to treat and also provide advice on choice of drug, when and how to adjust doses, and how to minimize non-adherence to therapy. The explicit intent of the guidelines is to assist physicians in achieving goal BP in the majority of their patients.

Two behaviors that could improve BP control are physicians' adherence to JNC-7 guidelines and patients' adherence to lifestyle recommendations and medication regimens. Several similarities exist between behavioral models that attempt to explain why physicians do or do not follow guidelines for preventive care [10, 11] and those that attempt to explain why patients do or do not adopt healthy lifestyle behaviors [12, 13]. For the physician, awareness (knowledge), motivation, and confidence (self-efficacy) all influence adherence to guidelines. For instance, physicians who do not know about or do not feel motivated to follow the JNC guidelines are not likely to prescribe them, nor are physicians who have the knowledge and motivation but lack confidence to follow the guidelines. Non-behavioral factors can also influence physicians' actions. These include external or systems-level factors such as time constraints, resources, and systems to prompt or remind physicians of the guidelines [10, 11, 14].

Similarly, behavioral factors may also influence patients' adherence to lifestyle recommendations and medications [12, 13]. Patients who are unaware that they need to change their behavior or who are not motivated will not change. In addition, external factors including food availability and environmental structures may affect how patients adhere to a healthy lifestyle. The Hypertension Improvement Project (HIP) behavioral interventions were designed to target the same psychosocial mediators and systems factors for both physicians and patients, presented graphically in Figure 1.

Strategies for implementing recommended practice to improve clinical outcomes include dissemination of printed educational materials (e.g., clinical practice guidelines), physician and patient training, feedback to physicians, dissemination of information linked to performance (e.g., audit and feedback, reminder systems), clinical performance measures (CPMs), and simple clinical bedside tools. Although randomized trials of these strategies are rare, there is evidence that such tools can change medical provider behavior and improve clinical outcomes [15‐17]. A major challenge is the integration of practice improvement strategies into the flow of clinical care; to accomplish this, health researchers and providers have borrowed techniques of Continuous Quality Improvement (CQI). It is the application of a CQI approach–applying tools shown to be effective–that forms the foundation for the HIP physician intervention.

The patient also plays a critical role in hypertension care. As noted above, the JNC guidelines include lifestyle modification for all patients with hypertension. Programs that result in successful lifestyle behavior change are generally based on Social Cognitive Theory [12] and techniques of behavioral self-management [18]. They are typically constructed using the Transtheoretical Model [13, 19] and motivational enhancement approaches [20, 21]. These approaches emphasize the importance of the individual's ability to regulate behavior by setting goals, developing specific behavior change plans, monitoring progress towards the goals, and attaining skills necessary to reach the goals. Self-efficacy (i.e., one's confidence in performing a given behavior) and outcome expectations (i.e., one's belief that changing behavior leads to a favorable outcome) are critical mediators of behavior change [12, 22]. An example of this approach can be found in the PREMIER trial of lifestyle interventions for blood pressure control–a multi-center trial sponsored by the National Heart, Lung, and Blood Institute that tested the effects on BP of 2 multi-component lifestyle interventions, relative to an advice-only control condition [23]. The PREMIER study demonstrated that a 6-month intensive behavioral intervention leads to healthy lifestyle changes, significant reductions in BP, and improved control of hypertension.

As is true for most chronic illnesses, hypertension management requires extensive efforts from physicians and patients alike, with support from the health care system. It was hypothesized that a quality improvement approach would maximize physician adherence to treatment guidelines, and a behavioral intervention would improve patient adherence to lifestyle recommendations. Both approaches may lead to significant improvements in BP control with consequent reductions in CVD risk. The HIP study offers a unique and innovative opportunity to test this hypothesis by determining, in a randomized controlled trial, the separate and combined effect on BP of a physician intervention and a patient intervention. The study design provides for an assessment of the effectiveness and cost of implementing the physician and patient interventions separately, thereby allowing health care organizations to make informed decisions about implementation of 1 or both interventions in the context of local resources. As participant enrollment concludes and data cleaning and analysis begin, this paper describes the protocol of the HIP trial and discusses the lessons learned thus far from designing and implementing the study, including issues encountered and subsequent changes to the study design.

The chronic care model suggests that BP control can be achieved by improving patient self-care and the systems through which care is delivered [29‐32]. Patient self-care efforts should be directed at counteracting the effects on BP of obesity, physical inactivity, poor dietary pattern, and non-adherence to prescribed medications [33]. Furthermore, efforts should be made to promote the use of quality improvement systems that can increase physician adherence to established clinical practice guidelines [34, 35]. In addition to improving BP control, these approaches can also reduce costs associated with hypertension and its consequences [29]. However, given the resources required to implement such approaches, it is critical that their effectiveness be rigorously established [36].

The Hypertension Improvement Project is a controlled trial to test the separate and combined effects on BP of a behavioral intervention for patients and a quality improvement intervention for physicians. The patient intervention employed proven behavioral methods for promoting healthy lifestyle and adherence to medication regimens. The physician intervention used a quality improvement approach to provide training, motivation, and feedback on performance in a non-threatening way to promote continuous self-improvement and adherence to clinical practice guidelines.

The physician intervention represents a form of quality improvement that can be described as "Facilitated Process Improvement" (FPI) [37]. Similar to CQI, FPI recognizes that a key barrier to implementing systematic improvement programs is provider time: time to become expert in and to apply CQI methodology and time to become familiar with the sorts of interventions that are likely to be effective in a given context. In FPI, several components of the CQI process are delegated to an expert team, with concurrent agreement by providers during education programs. In this case, the team identified the common limiting barriers to optimal outcomes–lack of easy access to guidelines and lack of ongoing feedback regarding conformance to guidelines (the latter issue may be termed "the difficulty of looking over one's own shoulder")–and created tools to address those barriers.

At the time at which this manuscript was prepared, the trial was in progress. All physicians and patients had been enrolled, and experience with the logistical aspects of study conduct had been garnered. There are numerous unique features and lessons gleaned thus far from the conduct of the trial that will set the stage for interpretation of the study results. More importantly, these features and lessons can inform other investigators planning similar studies.

Several of the challenges faced relate to the setting in which the trial is being conducted (Table 6). The study team considered it important to conduct the study in community clinics where such an approach to BP control would ultimately be implemented. Implementation of the HIP trial within a practice-based research network, while innovative, also required attention to issues that are not usually encountered within an academic medical center (AMC). The study required access to work space that was convenient and familiar to participants (i.e., in or near the clinic) but that had minimal impact on clinical care delivery. Not all medical clinics have space for study visits and group intervention sessions. The study team worked closely with clinic staff to schedule study visits on days when clinic rooms were available (e.g., provider day off) and used the waiting room (in the evenings), the clinic conference room, or a nearby facility for the group sessions.

Even though the study strived to cause minimal disruption to the usual clinic flow, the study team relied on the assistance of clinic personnel to arrange and facilitate several study processes. For example, clinic staff directed participants to the appropriate location for study visits, provided phlebotomy services, distributed blank CPM forms, and collected completed CPM forms. Each of these activities added to the staff's usual responsibilities, but several approaches helped to promote cooperation among study personnel and clinic personnel. First, the study goals and procedures were explained to clinic personnel prior to start-up of a new site. During enrollment and follow-up, the study team maintained frequent and open communication among study coordinators, investigators, and clinic management. Lastly, the study budget included modest compensation to the practices and to the enrolled physicians.

Travel from the AMC to the distant sites (ranging 10–40 miles) was not a trivial issue. The study team tried to travel as a group to minimize costs, and travel expenses were reimbursed. Rising fuel costs during the study affected the budget significantly.

Other challenges related to the choice of study population. The initial goal was to recruit low-income, ethnic minority participants, who are well-represented in the communities served by the participating clinic network. The study also aimed to comprise a population that was approximately 40% male, as men are often under-represented in lifestyle intervention studies. Because participants were recruited from the practices of enrolled MDs rather than the local community, previously effective targeted recruitment procedures could not be relied upon [38]. However, recruitment was targeted through access to the clinical database serving the participating practices, in which race, insurance, and sex data were available. Enrollment of low-income participants was encouraged by targeting recruitment efforts at patients in the database with no insurance, or with Medicaid or Medicare without supplemental insurance. Patient enrollment demographics were monitored periodically so that subsequent recruitment mailings could be prioritized to ensure a balanced study population. Midway through enrollment, the proportion of men was significantly below goal; in response, mailings and recruitment calls were directed preferentially to men. Despite these efforts, Table 2 indicates that most study participants were at least high school-educated and reported adequate financial resources, and fewer than 40% were men.

The patient intervention required weekly attendance at group sessions that lasted up to 2 hours. Several of the clinics served rural areas requiring significant travel by the participants. Some communities were not accustomed to the research activities of AMCs and, thus, patients in those areas may have been less inclined to join a research study for reasons related to lack of trust or information. While these factors mimic challenges to implementation of the patient intervention in real-world settings, they created a greater challenge for recruitment into a research study, especially given that recruitment was based on blind mailings. The initial recruitment letter asked potential participants to call the study team only if they were not interested in participating (an opt-out strategy); if the study team did not receive a response, then a follow-up call was made to answer any questions and determine interest in participating. However, the ratio of calls to enrollees was low, so the recruitment letter was changed to allow patients to also opt in. This option allowed participants to self-select for inclusion, which may have selected more motivated individuals but is still likely to generalize to a real-world implementation of the study intervention, which would ultimately require voluntary participation.

Recruitment of CHAs was somewhat challenging. The initial goal in enlisting CHAs was to recruit people with "natural helper" personalities from the same communities as the target patient population. To that end, flyers were posted in participating clinics, and clinic staffs and community leaders were asked to suggest people from within their communities. The response to this recruitment effort was relatively modest. Although we were able to recruit an adequate number of CHAs for the study with a minimal budget, our selection pool was limited. Future studies incorporating the CHA element may consider including a budget for a greater advertisement effort and other strategies to enhance recruitment.

Challenges involving the physicians were also encountered. A few providers left their participating clinical practices during the course of the 18-month intervention. In this situation, the provider's patients who were enrolled in HIP were referred to another participating provider within the practice. Also, completion of CPM forms often competed with busy and varying clinic schedules, the demands of patient care, or an upcoming review by the Joint Commission on the Accreditation of Healthcare Organizations. To minimize the additional work needed for collecting the CPM data, non-carbon copy paper was used, so that the top portion was used for the clinic note and the bottom portion was used for data entry. This was a brief 1-page form–the top portion was designed so that non-MD personnel (e.g., a triage nurse) could fill out the vital signs and medical history, while the bottom portion was filled out by the provider stating the medication changes and counseling that occurred during the visit. The physician strategy was developed by the study investigators who were not fully aware of all the QI initiatives (ongoing or planned) within the practices. The CPM form, while short and simple, was viewed by the practice staff as an additional form of which to keep track, and they had to modify their usual procedures 1 day a month to collect these data. With the increased use of electronic medical records, future studies should be able to integrate the collection of information about medication changes and counseling within routine charting, although this would also remove a reminder system that might contribute to improved BP management.

Developing the training modules was challenging due to the availability of multiple hypertension guidelines [8, 39‐41]. The original proposal was written when JNC-6 guidelines were widely used; after the proposal was written and before it was funded, the JNC-7 guidelines were released. The training materials were modified to match the newer version. To promote completion of the modules, they were offered on the internet (allowing providers to complete them when time was available), and providers were given CME credit to provide additional incentive for the time spent.

Overall, implementation of the study within a community setting was facilitated by including the network director (RJD) as part of the study team. The director had worked with the practices and providers on previous studies, and so her pre-existing relationships aided recruitment and troubleshooting of operational issues. Finally, it is important to define rewards and compensation for community practices, not only in terms of financial reimbursement, but also in terms of CME opportunities (categories 1 and 2), acknowledgment in publications, and dissemination of study results.

Hypertension affects approximately 29% of the adult U.S. population and remains a leading cause of heart disease, stroke, and kidney failure. Despite numerous effective treatments, only 53% of people with treated hypertension are at goal blood pressure. The HIP interventions offered an opportunity to significantly improve BP control and lower CVD risk in a broad cross-section of hypertensive patients from community primary care practices. The study design will allow an assessment of the effectiveness and cost of the physician and patient interventions separately, so that health care organizations may make informed decisions about implementation of 1 or both interventions in the context of local resources.