In prior years, one of the driving assumptions behind population health management technology was that it was important to focus on high-risk patients. The idea has been that fending off costly emergencies among these patients was the best way to control costs.
But if one market research report is on target, the practice of PHM may be evolving. According to a new study by analyst firm Frost & Sullivan, as a part of delivering value-based care, providers are deploying PHM applications addressing a far-broader range of constituencies, including not only evidence-based approaches to managing high-risk and at-risk populations, but also non-patient groups such as consumers, employees and physicians.
As PHM solutions grow more sophisticated, demand for such technologies is apparently increasing. Frost found that the US PHM market, which is valued at $3.1 billion this year, should hit $7.74 billion by 2022, a 149.6% jump in revenues. My assumption is that this will include a flood of new products that will offer new takes on PHM technology.
According to industry analyst Koustav Chatterjee, PHM vendors are using machine learning to do predictive disease modeling and deploying automated risk stratification to identify subgroups that can benefit from precisely-designed interventions. I imagine we’re talking about both pre-set and customizable filters here.
To achieve wider adoption of PHM solution, however, vendors will need to address data privacy concerns, Chatterjee said. That’s likely to stay a challenge as health IT leaders tap into multiple data sources with different security protocols in place, but these issues can be neutralized over time.
I don’t know if I agree with the short-term conclusions of this research. While I have heard anecdotes here and there about health organizations who are broadening their PHM programs, most still seem to be dedicated to the historic PHM models. In other words, they’re still focused on high-cost patients.
Sure, over time it probably makes sense to move beyond high-cost outliers and manage entire groups more effectively, but my sense is that most providers aren’t there yet; for example, only a minority of organizations have mature big data analytics programs in place, which would presumably be a prerequisite to the kind of segmentation Frost describes. I do suspect that PHM tools will begin incorporating AI capabilities soon, which should give such programs a boost, but they’re still at an early stage of development as well.
On the other hand, I think Chatterjee’s predictions should come to pass eventually, and that population health programs need to move beyond a focus on avoiding car crashes. It will just take several years before broader PHM capabilities go mainstream.