Top Data Analytics Tools for Population Health Management
A comprehensive population health management strategy requires health systems to leverage data integration, risk stratification, and predictive analytics tools.
Source: Getty Images
- Effective population health management requires collecting, processing, and analyzing large amounts of patient data, making data analytics tools key for any population health initiative.
Population health is often conceptualized based on a definition in a review published in the March 2003 issue of the American Journal of Public Health, which states that the term refers to “the health outcomes of a group of individuals, including the distribution of such outcomes within the group.”
The authors further posited that population health encompasses “health outcomes, patterns of health determinants, and policies and interventions that link these two.”
The American Hospital Association (AHA), as part of its Population Health Framework, indicates that leveraging robust data sources and advanced quantitative analytics solutions are crucial to understanding these patterns at the population level.
As of 2018, the AHA Population Health Survey shows that a majority of health systems are incorporating strategies to advance population health, but many still have work to do.
Below, HealthITAnalytics details the various data analytics tools that can be used alongside other solutions, like patient engagement technologies, for population health management. In addition, artificial intelligence (AI) and machine learning (ML) can, and often are, incorporated into all of the tools listed below. These technologies can support a vast number of population health-related use cases, including medication adherence, chronic disease management, mental healthcare, precision medicine, person-centered care for seniors, and flagging patient SDOH needs.
DATA INTEGRATION AND SHARING TECHNOLOGIES
Data integration involves combining data from multiple sources to create consolidated datasets. In healthcare, these data may be pulled from EHRs, patient-reported outcomes (PROs), disease registries, claims, geospatial information, demographic and socioeconomic information, social determinants of health (SDOH) information, public health surveillance systems, and other sources.
For population health management, data analytics, integration, and sharing are crucial to improving patient outcomes, addressing social determinants of population health, and bolstering care management. By creating high-quality, consolidated data repositories, stakeholders can better identify and define key populations, measure their care, and deliver care more effectively.
To this end, multiple health systems and other stakeholders have recently worked to improve population health-based data integration and sharing. In April, the World Health Organization (WHO) established its Health Inequality Data Repository (HIDR), the world’s largest repository of its kind to date, to help track health disparities and improve outcomes.
In 2020, the Indiana Network for Population Health was launched to support the secure exchange of SDOH data — including information on food access, transportation, and housing stability — to advance population health.
Providers and health systems that do not have a population health repository of their own can pursue other avenues to better understand their patient populations.
For instance, those beginning their journey will need to choose appropriate big data analytics tools, such as data visualization dashboards. Others who are further along may need to brush up on their data governance, information governance, data de-identification, and data normalization best practices.
By embracing data integration and sharing, healthcare organizations can turn their existing data into actionable clinical intelligence for value-based population health efforts.
PREDICTIVE ANALYTICS TOOLS
According to the Journal of the American Health Information Management Association (AHIMA), predictive modeling is a type of advanced analytics that can be utilized to forecast future health outcomes.
The benefits of predictive analytics in healthcare are broad, from providing clinical decision support to predicting patient mortality. For population health management, AHIMA notes that predictive analytics can help health systems track care trends like disease prevalence, comorbidities, and patient outcomes within their populations.
Other high-value use cases for predictive analytics include risk scoring for chronic conditions, preventing hospital readmissions, supporting precision medicine initiatives, and optimizing pharmaceutical outcomes.
Developing a successful predictive analytics tool requires that a healthcare organization have well-integrated data and a solid predictive analytics strategy. A successful strategy will help stakeholders identify appropriate use cases and support clinical staff throughout the tool’s implementation.
Risk scoring is also a key consideration for providers looking to incorporate predictive analytics into their population health management initiatives. The risk-scoring process allows health systems to better understand the needs and risk factors of a given population and improve care quality.
Some health systems that have implemented predictive analytics have seen significant successes.
Children’s of Alabama is leveraging the power of real-time analytics and predictive modeling in its cardiovascular intensive care unit (ICU) to forecast patient deterioration and extubation readiness. Doing so has allowed the hospital to pursue ICU Liberation for its population, an approach to improve care by freeing patients from ICU-related adverse outcomes.
However, healthcare organizations looking to leverage predictive analytics for population health management should know the potential risks and pitfalls. Researchers writing in the December 2019 issue of the Journal of the American Medical Informatics Association (JAMIA) argued that predictive analytics models should be made publicly available to promote transparency. This would enable independent external validation, assessment of performance heterogeneity over time and in various settings, and algorithm refinement.
“Hiding algorithms for commercial exploitation is unethical, because there is no possibility to assess whether algorithms work as advertised or to monitor when and how algorithms are updated,” the authors stated.
Insights from Deloitte about the value and risks of predictive analytics in healthcare highlight additional potential pitfalls: lack of regulation, algorithm bias, privacy concerns, the fast pace of technological change and its impacts on clinical decision-making, and moral hazard resulting from reliance on these tools.
Addressing these challenges will require healthcare organizations, federal regulators, and other stakeholders to collaborate on establishing best practices for using predictive analytics in healthcare.
RISK STRATIFICATION TECHNOLOGIES
Risk stratification and predictive analytics are often used in tandem to support population health efforts.
In healthcare, risk stratification refers to the process by which health systems systematically categorize patients based on their health status, in addition to other clinical, social, and behavioral factors. Doing so allows healthcare practices to engage in risk-stratified care management, in which patients are managed based on risk level to better anticipate their needs, proactively manage patient populations, and improve resource allocation.
Use cases for risk stratification tools in population health management vary significantly, with solutions targeting primary care populations, 30-day hospital mortality, patients’ medical spend and disease burden, and health activation.
Multiple health systems have developed their own approaches to tackle population-based risk stratification. Cleveland Clinic’s Risk Stratification Index (RSI) methodology and Johns Hopkins’ ACG System are two notable examples.
Other healthcare organizations have deployed risk stratification programs to address more specific use cases.
For example, Parkland Center for Clinical Innovation (PCCI) and Parkland Hospital in Dallas, Texas, have implemented a text message-based patient education program driven by a risk prediction model. The model uses SDOH data to determine clinical and population-level interventions to prevent preterm births.
The approach has allowed PCCI to better identify at-risk pregnant patients and enable early intervention. This has resulted in an 8 percent increase in prenatal doctor visits and up to 20 percent reductions in preterm delivery rates since the program’s 2018 launch.
To develop these population health risk scores and establish a population health management program, healthcare organizations must define their patient populations, determine the conditions and costs associated with these patients, and examine the risks involved.
Health systems can also choose a population health management company to support these efforts.
However, organizations building their population health management initiatives also need to look beyond risk scores, according to Johns Hopkins Medicine. This is necessary to ensure that patients are viewed as more than just a number, to bring attention to other factors that can influence overall health, and to compensate for any human errors that occur while leveraging the technology.
