Evaluating CommunityRx, a Self-Care Innovation for Older Adults, Using Agent-Based Modeling: Collecting and Analyzing Data to Build the Agent Based Model
Evaluating CommunityRx, a Self-Care Innovation for Older Adults, Using Agent-Based Modeling: Collecting and Analyzing Data to Build the Agent Based Model
CommunityRx is a health information technology-based innovation that, starting with the patient-health care provider encounter, facilitates self-care coordination for patients, caregivers, and providers. The CommunityRx database interfaces with electronic medical records to provide patients with a "HealtheRx." A HealtheRx is a list of community-based self-care resources tailored to the patients health needs (e.g., a person with diabetes receives information about podiatrists, nutrition classes, and other resources need to manage diabetes). CommunityRx aims to measurably improve health and health care while reducing health care costs especially in underserved health care settings. Specifically, the proposed research aims to 1) evaluate the impact of CommunityRx on health care utilization, cost, health, and patient-centered outcomes for program participants compared to controls; 2) examine the flow and spread of information to and through primary agents including: program participants, community health information experts, healthcare providers, and community-based service providers (businesses and organizations providing self-care resources); and 3) build and use an agent-based model to test the distributed impact, including economic effects, of CommunityRx system adoption on the demonstration area and predict performance over time by conducting experiments that vary assumptions about agent, environment, and population-level characteristics.
CommunityRx connects health care to self-care. CommunityRx begins with the medical encounter and functions like an e-prescribing system, modeled after medication e-prescribing.9 Rather than sending a medication prescription to the patient's pharmacy, a HealtheRx "prescription" is automatically generated at the point of care, delivered to the patient in paper (or electronic) form and stored in the Electronic Medical Record (EMR) for future reference. The HealtheRx is ontologically-generated, driven by standard demographic (eg. age, gender, address, language) data fields in the EMR and customized to patient problems/diagnoses like "homelessness," or "obesity." Like e-prescribing medicine, CommunityRx involves prescribing, fulfillment (patient signs up for a Weight Watchers class), and administration (patient participates in the weight loss program) (Fig. 2).10 The HealtheRx prescription is the primary information agent in the system, driven by simple ontologic rules that generate complexity, including iteration of the CommunityRx system itself, as human agents interact with the self-care information.
CommunityRx can be best understood as part of a complex adaptive system. The HealtheRx, in contrast to a drug prescription, is designed specifically to enhance the interpersonal aspect of the patient-physician encounter, informing both agents about self-care resources. A patient receives a HealtheRx and then contacts a Community Health Information Specialist (CHIS) or not. She seeks self-care resources or not.
Data captured by the CommunityRx system about patient referrals and needs are distributed to community-based service providers (CBSPs) in the form of quarterly reports delivered to CBSP contacts cultivated by CHIS. These data efficiently reveal to CBSPs real-time needs and gaps in self-care resources. The investigators hypothesize these data will be used by CBSPs over time to ensure supply. Youth who generate community resource data through employment with MAPSCorps also gain insight to community resources and engage with CBSP personnel and CHIS as they gather their data. Youth spread information to their networks about community resources and increase their use of these resources. MAPSCorps data about community resources are shared publicly (www.healtherx.org, www.southsidehealth.org, www.dondeesta.org). Self-care becomes more efficient. Supply meets demand. Patients and providers have more time and resources to devote to other salubrious activities and CBSPs become stronger. Over time, transparency in the market for self-care resources increases competition and quality.
This hypothetical dynamic is an example of emergent self-organization from a complex adaptive system. The intervention starts with a simple encounter, governed by simple rules, between patient and health care provider. As the number of these encounters grow, previously siloed sectors - health care and self-care - evolve a new kind of formation that is far more efficient for the community than the current state. CommunityRx drives this new formation through multiple agents who are unaware of the complexity they are producing: "…the self-organized structure simply emerges as a result of each individual doing their own thing, independently." In evaluating the economic effects of CommunityRx adoption, agent-based modeling (ABM) can test how close the attractor state, or end-point in a CommunityRx system configuration (set of assumed behaviors and designed interventions), comes to a Pareto efficient point of equilibrium. The effects of the CommunityRx intervention are non-linear, involve interactions and feedback loops, and therefore require a complex system modeling approach for evaluation.
A. Purpose or Hypothesis
Data inputs for agent based modeling (ABM) can come from a range of sources, including empirical quantitative and qualitative data, data from the literature, and expert opinion. Because CommunityRx targets people of all ages (0-99 yrs, to date), a prospective, experimental, community-based design (eg. RCT) to assess outcomes by age strata would be very informative about patient agents (the investigators expect age-group differences in behavior, social networks, and outcomes) but cost-prohibitive. ABM can accommodate assumptions made based on this important, but specific, population subgroup (or "testbed"), includes many agents, and allows for multiple simulations to assess the impact of variations in those assumptions for the much larger and more diverse population that the system-wide model includes. A cost-effectiveness analysis is needed to understand the true economic impact of CommunityRx on the total cost of the burden of disease. In addition, the research team brings clinical and research expertise and specialized interest in middle-age and older adult populations with chronic disease. Focusing on this subgroup builds on this track record and will meaningfully extend our contributions to the gerontology and geriatrics fields.
Specifically, the aims (and associated hypotheses) of this research include:
Aim 1. Evaluate the impact of CommunityRx on health care utilization, cost, health, and patient-centered outcomes for program participants (patients who receive care at the clinical demonstration sites and live in an 16 zip code area) compared to controls (patients who receive health care at the demonstration sites, but live outside the 16 zip code area), with a special focus on middle-age and older adults.
Aim 1a. Evaluate the impact of CommunityRx on health care utilization, health care costs, and on health outcomes for program participants (patients who receive care at the clinical demonstration sites and live in an 16 zip code area) compared to controls (patients who receive health care at the demonstration sites, but live outside the 11 zip code area) of all ages. NOTE: This aim is funded separately and registered on clinicaltrials.gov separately (see ID 1C1CMS330997).
Hypothesis: CommunityRx will decrease emergency/inpatient care utilization, decrease percent per beneficiary per year (%PBPY) costs and improve health.
Aim 1b. Evaluate the impact of the CommunityRx system on patient-centered outcomes in a randomized control trail of 200 program participants ages 45-74 and compared to 200 controls.
Hypothesis: CommunityRx will be associated with clinically meaningful improvements in: a) self-care behavior; b) perceived care quality; and c) quality of life.
Aim 1c. Characterize the economic value of care augmented with the CommunityRx system compared to usual care, based on the prospective participant-control study described in Aim 1b.
Hypothesis: Compared to usual care, care augmented with the CommunityRx system will be as cost-effective as commonly accepted medical devices and treatments.
Aim 2. Examine the flow and spread of information to and through primary agents including: program participants, community health information specialists, healthcare providers, and community-based service providers (businesses and organizations providing self-care resources).
Hypotheses: 1) Among the CBSPs receiving high volumes of CommunityRx referrals for people ≥45 years old (>1000/year), CommunityRx will produce a self-reported increase in: a) knowledge of community resources especially for older adults, b) referrals to other CBSPs, c) older client volume, and d) aging-related goods/services /programs inventory; and 2) Delivery of the CommunityRx intervention at the point of medical care produces knowledge about self-care resources in the community that spreads to secondary agents including members of patient and provider social networks.
Aim 3. Build and use an agent-based model to test the distributed impact, including economic effects, of CommunityRx adoption on the demonstration area and predict performance over time by conducting experiments that vary assumptions about agent, environment, and population-level characteristics.
Hypotheses: 1) The system-level value of CommunityRx is greater than the value quantified as %PBPY health care utilization savings and is projected to increase with population aging; 2) Experiments run on a systems-based model will predict and quantify the impact of strategies to optimize CommunityRx performance for improvements, sustainability, and spread to other settings; and 3) Systems-Based Modeling is an effective and efficient tool for large-scale evaluation of a health information technology-based intervention to improve health and health care.
Inclusion Criteria:
Exclusion Criteria: