Insurers Leveraging Patients' Personal Data for Predictive Analytics


Health insurers increasingly are turning to specialized marketing companies to obtain and analyze personal data about consumers, such as marital status and education levels, to help improve care and lower costs, the New York Times reports.


According to the Times, health care companies have long relied on data sets on patients' past health care consumption to estimate their future use of industry services. However, patients' growing use of websites and social networks to discuss medical information and the proliferation of fitness tracking applications has made additional data sources more widely available.

Meanwhile, the Affordable Care Act encourages hospitals and provider groups to reduce costs for certain patient populations.

Predictive Data Analytics

To meet those demands, more companies are incorporating individual consumers' household attributes and personal habits to refine their predictive analysis.

For example, the University of Pittsburgh Medical Center recently bolstered its predictive health analysis project by contracting with the marketing analytics company Acxiom, which uses public and private sources to glean information about consumers' behaviors. Specifically, UPMC incorporated data on patients':

  • Education level;
  • Household income;
  • Marital status;
  • Number of children at home;
  • Number of cars;
  • Race or ethnicity; and
  • Other information.

After adding the data, UPMC found that mail-order shoppers and people who used the Internet were more likely than other beneficiaries to use more emergency care services.

Although UPMC said that it has not yet used its household data findings, it does place patients into different "market baskets" based on more traditional health data to help link care coordinators with patients flagged as high-risk.

Pamela Peele, chief analytics officer for UPMC's insurance services division, said that the system helps UPMC develop "better prediction models and allocate [UPMC's] clinical resources," adding, "If you are going to decrease the costs and improve the quality of care, you have to do something different."

Similar collaborations include:

  • An East Coast health insurer that enlisted Predilytics -- a Burlington, Mass.-based health care analytics company that focuses on socioeconomic, demographic and consumer buying information to identify high-risk patients -- and discovered that patients who were unable to arrange transportation or schedule timely appointments were more likely to be hospitalized; and
  • MedSeek, a Birmingham, Ala.-based software and analytics company, that offers services to help hospitals "virtually influence" patients' purchasing behaviors by targeting marketing pitches to consumers' age, family status, gender, income, race, religious beliefs and risk assessment.

Patient Privacy Concerns

However, some technology and health law experts have raised concerns about patient awareness and privacy.

For example, Frank Pasquale, a privacy and health law expert and a professor at the University of Maryland School of Law, voiced concerns about how patients are often unaware that such data are being collected, calling the practice, "intensive" and "intrusive." He also noted that the analysis could lead to "differential treatment of customers," which he argued should be disclosed in the medical industry.

Anita Allen, a privacy law expert and the vice provost for faculty at the University of Pennsylvania, said, "I think we should be careful of running gung-ho into an area of health care analytics that may disadvantage deserving patients."

According to the Times, at least one marketing firm -- Acxiom -- has admitted that their data can be outdated or incorrect.

Pasquale suggested that hospitals and insurers be required to make the data public to ensure that the information is used equitably, adding that while it "would be great if these programs do work," they are "kind of untested right now" (Singer, New York Times, 6/30).

to share your thoughts on this article.