Wednesday, October 2, 2013
Hot Spotting - Targeting Healthcare Frequent Fliers
The Economist article, 'A Different Game', is relevant because it addresses a need for deep analysis of inert data to extract actionable information. Actionable information is referred to as 'intelligence' because it provides a tangible blueprint to address significant, but hidden problems.
This novel approach to understanding hidden problems that are lurking with in large populations, through data analysis, is at the heart of cost reduction efforts within the healthcare industry. Healthcare data is generated by a multitude of sources, including patients records, claims reports, hospital equipment, mobile devices and insurance reports. This data contains hidden messages to address the cost and quality issues that plague healthcare.
One interesting use of data mining techniques in healthcare has been to analyze population trends to identify low cost solutions to achieve high margin gains. This represents the starting point in the healthcare supply chain.
The significance of this area of research reveals itself in a New Yorker article entitled 'Hot Spotters' by A. Gwande. In Hot Spotters, the significant finding and relevant issue to our study of information technology is that data analysis of public health records has revealed a few individuals within certain geographic areas that account for the most healthcare spend. The key is using data analysis to identify these 'frequent fliers' or high spend individuals. Again, this identification of high risk individuals is the start of the healthcare supply chain. Later steps include allocating resources and care models.
In Camden, New Jersey, “one percent of the hundred thousand people who made use of Camden’s medical facilities account for thirty five percent of its costs.” These ‘frequent fliers’ showed higher rates of hospital admissions, emergency room visits and complex chronic conditions. Providing intensive outpatient care to communities where a relatively low number of high needs patients generate the costliest medical bills will produce high value results.
These high cost individuals could only be identified through sophisticated data mining and GIS techniques. Health IT, data analytics and geographic information systems identified specific hot spot apartment buildings. Two buildings in Camden, NJ accounted for $200 million dollars in health-care bills, and the most expensive patient cost insurers $3.5 million. These buildings were identified using patient records, GIS and data analytics tools. The relevance to our study is that this analysis of inert data led to 'intelligence' or actionable items and created a blueprint for action throughout the healthcare supply chain.
In light of the effective nature of data analytics to predict trends in population health and target individuals who create healthcare costs outside the normal distribution, are there privacy issues that need to be addressed? Should we implement policies that allocate resources towards highlighting high crime neighborhoods and sick neighborhoods that incur disproportionate costs or is this a negative profiling technique? Can we justify these invasive measures by explaining that this has the highest marginal benefit to society?