Data Quality and Reporting Guide

Organizations interested in leading-edge developments in data quality and reporting should consider the implications of new technology for both collecting and analyzing data. As recognition grows about the impact of home and community life on patient health, so does interest in supporting patients to monitor and improve health in their home environments.

Remote patient monitoring (RPM) is a form of telehealth that uses technology to enable patients to collect and share physiological data, such as blood pressure and blood glucose levels, and behavioral data, such as medication adherence and physical activity, from wherever they are.^[1] This type of remote patient monitoring can improve care management programs for chronic conditions and enable healthcare teams to intervene early when a health problem is beginning to escalate for a participating patient. Systematic reviews show promising results for the effectiveness of remote patient monitoring for supporting patients with common chronic conditions, such as diabetes, hypertension, chronic obstructive pulmonary disorder and cardiovascular disease, such as congestive heart failure. ^[2]

Remote patient monitoring outcomes can vary and key elements to optimize effectiveness include:

• Accurately identifying and engaging populations at high risk, especially those with high or rising risk for hospitalization.
• Effectively identifying a decline in health, which can be more difficult for some conditions than others. Even in the cases where acute changes to health status are hard to detect, RPM can be used to monitor longitudinal progression of illness.
• Ensuring timely and responsive care by creating automated or user-friendly data sharing processes for patients, and ensuring that care teams are providing ongoing monitoring and rapid responses with bidirectional communication for any issues that arise.
• Providing personalized care, including collaboratively designing RPM approaches with patients, offering personal and repeated patient training on RPM devices, as well as establishing personal thresholds for alerts to the monitoring team, which may vary from patient to team.
• Combining RPM with self-management support, including patient education and coaching to improve knowledge, and skills and behaviors to improve self-monitoring and management.
• Offering collaborative and coordinated care, which includes both patients and families or caregivers, multidisciplinary representation from the patient’s care team in both primary care and specialty care, and proactive facilitation of not just monitoring, but enhanced partnership between patients and providers.

RPM is likely to be most effective when understood as a new model of care, not just a new technology. This understanding requires thinking carefully about collaboratively designing the approach with all stakeholders, implementing and refining workflows, committing adequate resources to patient education and self-management support, as well as ongoing, bidirectional communication between patients and monitoring teams.^[3] From an equity standpoint, while RPM and other forms of telehealth may be useful to patients who have barriers to accessing clinic-based care, there may be additional obstacles to engaging diverse patients in RPM effectively, from low digital or health literacy challenges to language and cost barriers. Culturally congruent or contextualized care with wraparound support for RPM, such as community health workers supporting patient adoption, may mitigate some of these challenges.^[4]

Although medical professionals have been looking for ways to leverage artificial intelligence (AI) since the 1950s, advances to AI in healthcare have exploded in recent years with enormous growth in both technologies and research.^[5] Under the umbrella descriptor of artificial intelligence, there exist a range of technologies with varying relevance to enhanced data quality and reporting for population health.^[6] Although AI has a very broad range of implications for both public health and healthcare delivery, potential applications for primary care data quality and reporting include integrating AI into remote patient monitoring, risk prediction and intervention, and risk-adjusted empanelment.^[7] The most relevant and immediate application of AI for population health in community health settings may be enhancing risk identification and prediction using medical, behavioral and social health data to create a more holistic picture of both individual and population-level risk characteristics.^[8][9]

Although AI holds promise for enhancing healthcare delivery and population health, the technology also has the potential to reinforce and worsen health disparities. AI technologies that rely on existing datasets to learn and build functionalities will recapitulate biases that are present in the initial data, and may then be capable of replicating such biases at enormous scale.^[10] Organizations developing or investing in AI should be careful to consider health equity implications for new technologies, as well as advocating for regulatory standards to mitigate the risk of reinforcing existing biases.^[11]