Population health management is a hot topic in 2020, but what does it mean and why does it matter? Defining population health management is challenging—there is no clear or standard definition of population health within the healthcare industry. If you ask 10 different people to define population health, you will most likely get 10 different answers. Some organizations have a broad definition whereas others have a very focused idea of what population health management means for them.
For our purposes here, let’s define population health management as the aggregation and enrichment of disparate data sources and the analysis of that data to inform quality initiatives aimed at improving health outcomes and reducing care costs. With a data- and analytic-driven approach, organizations can better understand their patient population and make informed strategic decisions based on predictive analytics that identify, predict and prioritize at-risk populations, creating the foundation for the optimal intervention strategies.
Population health management initiatives hold great promise for improving outcomes, but there are some common organizational challenges, including inaccurate and incomplete data, lack of a fully integrated population health tool driven by real-time data, lack of insight to address social determinants of health, lack of resources for effective care management programs, and care management programs that don’t address the clinical and non-clinical needs of each patient.
The components of an effective population health analytics tool include:
- Aggregation and integration of disparate data such as clinical and consumer data, social determinants of health data and national and regional benchmarks
- Population segmentation and risk stratification to provide a more holistic view of patient care informed by clinical and non-clinical data elements
- Predictive analytics to identify, predict and prioritize at-risk populations to inform optimal intervention strategies
- Flexible, advanced analytic platform that can inform and/or integrate with existing strategic programs such as case management software
Q&A with Johns Hopkins Medicine
To learn more about population health management and explore developments in the population health space, Inovalon sat down with Harriet Martyn of the Population Health Analytics team at Johns Hopkins Medicine.
Q: Population health management is such a large concept. How would you make it a more tangible and actionable methodology?
A: The idea behind population health management is that once you understand a population’s health profile, those insights can inform actions to improve the outcomes of large groups of individuals with common health needs. There are many steps to a population health management process, which can be tailored to individual regions or health plans, but some common steps include:
- Analysis of clinical and social data to understand population-level risks
- Segmenting the population according to its needs
- Creating a staffing infrastructure and network design that supports interventions
- Implementing disease management interventions based on population needs
- Measuring the effect of interventions through population health outcomes, cost savings and quality improvement metrics
Q: What kind of population health analytics tool is best for population health management? What data sources should be employed for population health management?
A: Choosing the right kind of population health analytics tool depends on the problem you are trying to solve. Chronic diseases are now the primary driver of healthcare expenditure, and populations are ageing as well as living longer. As a result, population health managers are now more interested in understanding the patterns of co-morbidity that are driving health risks and their impact on life expectancy. Since chronic diseases are not time limited in their nature, taking a whole-person approach to analyzing risk is important in determining the right treatment plan for an individual. Performing risk adjustment that accounts for age, sex and morbidity can further help to stratify your population based on levels of risk. In addition to analyzing clinical data to understand health risks, social and economic factors have been shown to account for 40% of health outcomes. Multiple data sources, both clinical and non-clinical are important to gain a comprehensive picture as part of your population health management approach.
Q: How is the Johns Hopkins’ ACG® System — a model that predicts an individual’s health over time using existing data from medical claims, electronic medical records and demographics — used for population health management? Do you have any real-world examples?
A: The Johns Hopkins ACG System is used by commercial and public payers, health systems, integrated delivery networks, and governments in the United States, as well as by customers in approximately 20 countries.
Closer to home, Johns Hopkins HealthCare uses the ACG System to support our four health plans. They also used the ACG System to support the Johns Hopkins Community Health Partnership to identify high users of healthcare resources based on their patterns of co-morbidity, diagnoses of substance abuse or mental health, and a high risk of future hospitalization based on their prior inpatient use and clinical conditions. Social risk factors for barriers to care were also assessed. The resulting target population was individuals who are dual-eligible, living around the Johns Hopkins East Baltimore campus who had heart failure, COPD, sepsis, mental health and addiction disorders. Their main barriers to care included transportation constraints, inability to obtain needed eyewear, and inadequate emotional support. As a result of the risk analysis, the Johns Hopkins Population Health team worked with two local community organizations to implement neighborhood navigator programs, integrated care teams, home-based primary care programs and patient engagement training. As a result of the interventions, the rate of hospitalizations, ED visits and readmissions decreased, and total cost of care decreased for Medicaid beneficiaries over a 12-month period. Patients also reported a high quality of care, and a good experience with healthcare practitioners who communicated clearly to them and listened to their concerns.
Q: How do you see population health management evolving over the next five years?
A: There are a lot of buzz words in population health management today, and especially with regard to the population health analytics tools available, from machine learning to consumer data, to natural language processing. Sometimes it is hard to make sense of how the tool will help you, especially if you are just starting out on your population health management journey.
Social determinants of health data will likely add a lot of power to analytic models and help predict risk beyond just clinical data.
Machine learning is a broad topic, and it’s important to understand that there are many algorithms that support different applications. Machine learning models can certainly improve predictive power, but it’s important to understand what dataset they have been trained on, whether it is clinically cogent, and how transparent the methodology is to be able to explain this to your stakeholders – be that a primary care provider or a CFO.
Natural language processing tools will give us further ability to mine unstructured data from the EHR. For certain conditions, like geriatric syndromes, including unstructured EHR data, can identify between 1.5 and 7.5 times as many cases than structured claims and EHR data combined.
As we continue to see a shift from volume to value-based payments and providers in the United States are being incentivized to focus on health outcome measures, the need for population health analytic approaches will become integral to identifying individuals with the highest healthcare needs and prioritizing care based on risk.
The Value of Population Health
Looking at population health management broadly, the benefits are vast and positively affect payers, providers, individual patients and the community. Open communication and data sharing better aligns payers and providers to manage and care for their populations, improving health outcomes at lower costs. For those receiving care, the shift from reactive to proactive care models promote holistic care and healthy behaviors, empowering people to actively manage their own health. With the growing adoption of IT in healthcare promoting innovation, the possibilities are endless for improving healthcare.