Healthcare's clinical community should view population health as a frontier for deepening understanding about the nature of disease, providing incisive clues about patient demographics, behavior patterns and physiological makeup. Large group samples hold the potential to unlock mysteries that physicians may not even realize exist, says Jonathan Teich, MD, chief medical information officer for Amsterdam-based Elsevier.
"It's a paradox," he says. "By studying large populations we can do more precise things."
The study of population health is integral to medical learning, says Teich, also an assistant professor of medicine at Harvard and board-certified attending physician in emergency medicine at Brigham and Women's Hospital, because it provides insights into behavioral issues, new dimensions of chronic diseases and precise clinical data that can be used to make healthcare more cost efficient and clinically effective.
"Population health is something we can and must do," he says. "We currently don't have a cost effective healthcare system and even though medicine has become very advanced, we still do a lot of guessing. We need a lot more information - we can't keep relying on anecdotes. We need data from large populations to help us further our knowledge," Teich adds.
With chronic diseases and hospital readmissions caused by their complications representing the biggest cost drivers in healthcare, the medical community needs to concentrate on better managing disorders such as obesity, diabetes, CHF, COPD and asthma, Teich says. Using large population samples by which to apply various computational metrics, he says physicians can drill down into specifics they never knew existed.
"The sweet spot is where there are variations and unknowns," he says. "While we know how to treat COPD patients, there may be answers contained in populations about how the disease affects certain demographics. Using a large population lets you mine for that information. You can ask a lot of questions of that data and find something you weren't looking for."
At Elsevier, Teich and his team have been focused on retrieving the knowledge that comes from population data. Genomic data could yield revelations about cancer and other serious diseases, he says. Moreover, Teich says mining texts could also be a breakthrough because "there is a lot of information in texts that has been bottled up and hasn't been used in computation."
Deciphering coding
Although only 20 percent of medical content is currently coded and usable for quality improvement, Dan Riskin, MD, CEO of Menlo Park, Calif.-based Health Fidelity has been working diligently on the text front to extract more data.
"As we move into an era of data-driven healthcare, information will increasingly be extracted in an automated fashion," says Riskin, consulting assistant professor of surgery at Stanford University. "Automated data extraction was initially deployed for billing, but is increasingly being used to empower quality analytics and population-based approaches. The process of extracting information can be as simple as text matching or as complex as natural language processing (NLP). To date, NLP has yielded the highest accuracy for automated extraction of information in healthcare."
NLP has been used to extract a broad array of quality measures, such as smoking, heart disease and diabetes, Riskin says, and in the process it has enabled quality improvement efforts, bypassing or enhancing the manual data collection processes currently in place.
"While efforts to date have been limited in scope, it is clear that quality improvement will be more powerful when the manual system is enhanced or replaced with automated processes," he says.
Going forward, Riskin is confident NLP has the potential to transform quality improvement and population-based health. Current national initiatives typically focus on 10 to 15 quality measures out of the hundreds known, limited by manual processes and the expense of extracting information.
"As the system becomes automated, healthcare organizations can focus on the full breadth of quality measures rather than just a handful," he adds.
Series of ovals
Jonathan P. Weiner, director of the Center for Population Health Information Technology at Johns Hopkins Bloomberg School of Public Health, sees population health as a series of ovals that emanate outward from the physician-patient relationship.
In a diagram Weiner constructed to illustrate the expansive impact of population health in a digitized environment, ovals cascade wider from the original episode of care. Immediately beyond the physician-patient encounter are the patient's family and the physician's practice team. Further out from the family is the community and outside the practice team is the integrated delivery network, accountable care organization and virtual network. Surrounding all of this are the IT mechanisms - electronic health records, claims information systems, computerized physician order entry, clinical decision support, social media and others.
"It's not just about 20-minute doctor-patient interactions, but the whole life around it," says Weiner, who also serves as professor of health policy and management and health informatics at Johns Hopkins. "Our mission is public health. What we hope to do at our school is to cover the entire milieu - the health system and community - and not just on the physician, who tends to focus on one patient at a time."
Because public health relates to the big picture, Weiner has a concrete perspective on how all facets of health and community fit together. Admittedly, it is a complex IT puzzle that is still in the midst of being assembled, he adds.
"Johns Hopkins is one of the only institutions in the world that blends public health, medical, nursing and health systems together," he says. "Our goal is to be an agent of change and through evidence generation we hope to transform healthcare while being honest brokers about it."
