It doesn’t seem like a day goes by without the mention of health care on television, in the newspaper, or online. The passing of the 2009 Patient Protection and Affordable Care Act is creating a large amount of disruption in the U.S. health care sector. Couple that disruption with the Health Information Technology for Economic and Clinical Health Act, and you have a data revolution that is still in its infancy aimed toward health care management.
Predictive analytics is the application of data analysis and modeling with historical data to make future predictions for a given incident or set of events. How is predictive analytics being applied to the health care sector today, and is it having an impact on cost? To better understand this situation, let’s look at one example in depth.
An estimated 125,000 annual premature deaths result from patients not properly complying with their prescription medications; an annual cost of $290 billion comes from avoidable medical care. Examining noncompliant patients taking medication for hypertension, each patient on average spends an additional $4,000 annually compared with a compliant patient. By identifying those patients in advance who are at high risk for noncompliance, doctors can reduce the increased health care costs of those patients down the road. But, how do you identify these high-risk patients?
An interesting approach that is being used to calculate such predictive behavior is a patient’s FICO Medication Adherence Score. In this example, analysts pull readily available data for a patient, such as age, gender, martial status, length of time in the home, and geographic region, and have found that these factors have a high correlation with the chance that a patient will be noncompliant with his or her prescriptions. A risk score is then calculated for the patient, and based on this score the doctor can take preemptive measures to decrease their risk of noncompliance, such as:
- Reminding patients via phone, SMS, or e-mail to take their medication. Limiting the number of medications a patient needs to remember to take, where possible. Educating patients and consumers on alternative financial means to pay for their prescriptions. Ensuring that the patient can read the prescription in his or her own language (Spanish, French, etc.); and having an outside assistant physically ensure that the patient is taking his or her medication.
Unfortunately, several impediments are slowing the application of predictive analytics to health care, including:
- Numerous disparate information sources to corral;
- Ensuring the quality of the data being analyzed;
- Lack of human resources with the appropriate skill sets;
- Providing needed information when it’s most impactful;
- No one in the organization being charged with leading a data-centric effort;
- The organization itself not seeing the value of data analytics in its strategy; and
- Much of the data collection still being a manual process.
How familiar do the issues above look compared with your own industry or organization? These problems are not new, even if you don’t work in health care. Being able to make smart, well-educated and informed decisions largely depends on the quality of data possessed and the ability to analyze the data effectively. Many organizations want to be data driven, but getting there is not easy, and predictive analytics within health care is going to be a growth industry for decades to come.