Enhanced care planning and clinical-community linkages versus usual care to address basic needs of patients with multiple chronic conditions: a clinician-level randomized controlled trial

Physical inactivity, poor nutrition, tobacco use, and overconsumption of alcohol cause much of the illness, suffering, and early death related to MCC [18]. Unhealthy behaviors are ubiquitous in the USA—only 21% of adults exercise adequately, 40% eat insufficient vegetables and fruit, 15% smoke, and 20% binge drink an average of four times per month [19‐22]. Unhealthy behaviors increase cardiovascular risks, cause elevated blood glucose, exacerbate diabetes-related complications, cause weight gain, and impair mood, ultimately leading to uncontrolled chronic conditions. Given the harms of unhealthy behaviors and the benefits of behavioral counseling, the US Preventive Services Task Force (USPSTF) recommends screening and counseling for healthful diet, physical activity, obesity, smoking, and alcohol misuse [23‐27]. Mental illnesses and psychosocial distresses are both chronic conditions and—as in the case of stress and anxiety—are contributors to poor MCC control. Distressed patients are less likely to seek medical care, adhere to care plans, or maintain healthy behaviors. Likewise, patients with MCC have lower physical and social functional status, less favorable mental well-being and health perceptions, and greater pain than do patients without chronic conditions [28]. Mental illness, psychosocial distress, and uncontrolled MCC reinforce and exacerbate each other [29]. Social needs—defined here as housing, food, transportation, finances, employment, education, and safety—also influence health outcomes [30‐36]. They contribute to health inequities, higher costs, and overutilization of health services. Clear evidence shows that social determinants of health have a greater impact on morbidity, mortality, and quality of life than medical care itself [30, 34‐42]. The National Academy of Medicine (NAM), the Centers for Medicare & Medicaid Services (CMS), and the World Health Organization have called for primary care and public health to address social needs [35, 43‐45].

The chronic care model informs how communities and healthcare systems can better address MCC [46‐49]. It describes how self-management support, delivery system design, decision support, and clinical information systems can lead to a prepared, proactive practice team and an informed and activated patient (the latter specified as a required element to achieving improved health outcomes). While educational programs and efforts by clinicians to engage patients can help produce an “informed and activated patient,” patients cannot participate fully in their health care if they are struggling to meet basic needs. Patients facing these challenges are likely less capable of participating in a higher-level functional activity, such as health care (Fig. 1).

Effectively addressing health behaviors, mental health, and social needs can be more challenging than traditional medical services. Health behavior counseling requires dozens of hours of face-to-face contact by a multidisciplinary team over many months [50, 51]. The mental healthcare delivery system is not well integrated with medical care [52]. The healthcare system is only beginning to consider its role in addressing the complex problems patients face in getting stable housing, food security, employment, loneliness, and education [53]. Social needs cover a wide range of problems, each involving different sectors, agencies, and networks of community organizations. Community programs, social service agencies, public health departments, policymakers, and other stakeholders have been working arduously to address these needs. For example, many community recreation centers offer diet and exercise programs, nutritional counseling, group classes, and chronic disease programs. Policymakers are considering “health in all” programs, redesigning public housing, enacting laws to curb smoking and unhealthy diet, and supporting social services and mental health. Health departments, community service boards, housing authorities, food shelters, and other entities are addressing poverty, job training, unemployment assistance, transportation services, violence prevention, and advocacy. These activities are generally outside the purview of traditional primary care training and activities. An unmet need is for primary care clinicians to find their place among those efforts—a pathway for clinicians to contribute.

Primary care clinicians have need for systems to connect patients with these resources because of the frequency with which they encounter behavioral, mental health, and social needs in routine practice. Virginia Commonwealth University (VCU) led a national study to field My Own Health Report (MOHR), a health risk assessment and care planning tool administered in nine diverse settings to screen for 13 health behavior and mental health needs. Patients reported an average of 5.8 needs [54‐56]. The majority had six or more needs, and less than 1% had no needs. In an ongoing study asking patients about social needs in 12 practices that serve an affluent Northern Virginia population, high rates of basic social needs were reported—71% had at least one [57, 58]. Although addressing every need is not feasible, 3–20% of patients wanted clinician help with a limited set of prioritized needs. A patient-centered approached guided by patient interest in addressing needs is more manageable and likely to help patients.

Health systems are increasingly held accountable for health outcomes and reducing costs. Recognizing the importance of social determinants of health, health systems are developing models for connecting patients with programs addressing behavioral health and social needs. Hospitals, emergency departments, and other providers are partnering with social services, public health, and community-based organizations. Models are proliferating across the country, including Hennepin Health, the Chicago-based Community Rx, the Michigan Public Health Institute Pathways to Better Health Community Hub, the American Hospital Association, Practical Playbook, and the EveryONE Project [45, 59‐61]. These projects show the importance of cross-sector partnerships, data systems to bridge clinical and community care, and the need for community health workers and navigators [45]. In 2017, the CMS catalyzed the movement by funding 31 communities to develop and test new payment and service models, termed Accountable Health Communities (AHCs), which build clinical-community partnerships [62]. Private payers, employers, and the public health sector are further supporting these efforts, resulting in growing national momentum for change.

While these changes are exciting, they have two major limitations. First, few are being evaluated in randomized controlled trials (RCTs). The CMS AHC program utilizes a dissemination design, focused on maximizing the number of participants and discouraging comparison groups. This limits their ability to demonstrate efficacy. Second, few have a defined role for primary care; most are designed for use by health systems in centralized care settings, such as emergency departments or hospitals. This hospital-centric focus is reflected in the CMS-mandated outcome of AHCs: reduction in hospital and emergency department admissions.

Primary care practices, in contrast, are distributed across communities and need innovative models for tapping into such systems. Few primary care practices can build their own infrastructure. An innovative solution is to establish practice systems that take advantage of the hospital- and health system-based programs that are proliferating across the country. To do this, primary care practices need (1) a system to help patients relay their needs and collaboratively create care plans that incorporate their values, preferences, and personal, social, and clinical context, and (2) a system to connect patients with community resources to help patients achieve their care plan goals.

Beginning December 2019, we will recruit clinicians from Virginia Ambulatory Care Outcomes Research Network (ACORN) primary care practices located in the Greater Richmond Region. ACORN has academic affiliations with nearly 500 primary care practices, 53 of which are located in the Greater Richmond Region. Practices range in size from 2 to 18 providers and operate under diverse ownership and insurance models.

Sixty clinicians will be randomly and equally (1:1) allocated between the intervention and control conditions using stratified randomization with matching achieved by randomly allocating clinicians within each of the four cross-classifications of age (less than 50, 50 or older) and sex (male or female). Allocation will be conducted by the study biostatistician, who will use the R statistical software to generate random numbers (between 0 and 1) for each clinician, allocating to the intervention for numbers greater than or equal to 0.5, and otherwise allocating to the control. Patients and chart auditors will be blinded to allocation, though clinicians will not be blinded due to the nature of the intervention. From the EHR, we will obtain a list of all patients seen the prior year including their demographic information (age, sex, race, ethnicity, insurance type), diagnoses, and vital signs (body mass index [BMI] or height and weight). After identifying all eligible patients, we will use stratified sampling to randomly sample approximately 50% white and at least 30% black, 10% Asian, 10% multiracial, and 10% Hispanic patients (of any race); we will purposefully oversample Asian and Hispanic patients to ensure representative samples. Eligibility criteria are age 18 and older and the presence of two or more chronic conditions (cardiovascular disease, hypertension, hyperlipidemia, diabetes, obesity, MDD).

In March 2020, we will randomly select 50 patients from each intervention and control clinician’s panel and mail a baseline survey. We will review the charts of the survey respondents and assess their survey responses to identify if their MCC are uncontrolled. From this sample, we will randomly select 10 patients with at least one poorly controlled MCC for study inclusion. Study research coordinators will contact patients to arrange an in-person or virtual meeting to consent the patients, enroll them in the study, and create the patient’s initial MOHR account. On the consent form, participants will be asked if they agree for their data to remain part of the study (unless requested otherwise) should they choose to withdraw from the trial. Participants will also be asked for permission for the research team to share relevant data with the practices taking part in the research or from regulatory authorities, when relevant. The consent form states that information collected as part of the study will not be used or distributed for future research studies. This trial does not involve collecting biological specimens for storage.

Clinicians and patients can withdraw from the study at any time. Leaving the study will not affect ongoing medical care. There will be no specific clinical or social criteria for discontinuing or modifying allocated interventions.

To address the questions in our three specific aims, we will use five data sources: MOHR, chart review, patient navigator and CHW field notes, patient postal survey, and patient semi-structured interviews.

An overview of our data collection methods is presented in Table 2. Outcomes span three domains: implementation, effectiveness, and context. These outcomes will provide an understanding of the intervention, capacity of primary care and communities to do care planning and participate in the intervention, contextual factors that influence outcomes, and potential for dissemination.

The following power calculations are based on Murray (1998) and Donner and Klar (2000) [95, 96] to account for varying treatment effectiveness between clinicians due to (1) clinician-based randomization and (2) nesting of patients within clinicians. To account for not being able to measure 6-month and 2-year outcomes in some patients (e.g., because they move, change practices, etc.), we aim to enroll 10 patients from each of 2 clinicians from 30 practices, and conclude with data on 8 patients per clinician, for a total of 480 patients. Assuming an intra-class correlation (ICC) of 0.05, a 5% type I error rate, and a between-patient standard deviation in the number of uncontrolled MCCs of 1.0 (aim 2), we will have sufficient power. This provides 80% power to declare a 0.26 decrease in the mean number of uncontrolled MCC in the intervention group significantly different from a 0-unit change in the control group. The 0.26 decrease in mean number of uncontrolled MCC correlates with the hypothesized 10% improvement in aim 2. This goal was selected as a clinically meaningful improvement for patients.

The overall study timeline is shown in Fig. 4. Funded in March 2019 by the AHRQ (1R01HS026223-01), preparation for recruitment has begun. The study is expected to begin recruitment of clinicians in August 2020 and patients in October 2020. Recruitment will continue over a 2-year period and be completed in August 2022. A data safety monitoring board (DSMB) will be assembled at the start of clinician recruitment and will meet annually to perform an annual audit that will include monitoring study progress, participant safety, and interim results. If the DSMB has any study concerns, they will be able to terminate the study early. The DSMB will include a biostatistician, a primary care clinician researcher, and a social worker. The DSMB will be an independent body free from competing interests. Patients, clinicians, patient navigators, and the CHW will be able to report adverse events to the investigators or the VCU IRB at any point. All reported adverse events will be shared with the VCU IRB and the DSMB. Any protocol modifications will be updated on ClinicalTrials.gov and submitted to the VCU IRB.

Findings from this study will be published in medical journals and presented at medical conferences. Participating clinicians and patients will be given summaries of results prior to publication, and feedback will be incorporated in publications. The investigators will have no restrictions in what they can publish or present. This article presents protocol version 1.2 of the study, which was finalized September 15, 2019.