USC News

Menu Search

Can big data lead to lower costs for health care?

USC Viterbi alum uses data science to determine possible causes for rising costs of prescriptions

Medicare’s drug prescription data for 2013 accounted for $103 billion. (Photo/Laura Gilmore)

How many brand-name psychotherapeutic drugs are being prescribed in the United States in lieu of their generic forms?

Data analysis could play a key role in answering this and other questions that could help policymakers improve the American health care system.

USC Viterbi School of Engineering alum Sundeep Pattem has taken an initial step by studying public health care data to determine where there are opportunities for the government to cut costs. His latest research looks at Medicare’s drug prescription data for 2013, which, he said, accounted for $103 billion.

By studying how many times a drug was prescribed, how much each prescription costs and considering statewide comparisons by drug class and generic versus brand-name prescription rates, Pattem was able to detect several trends in the data and possible causes for rising costs.

“Currently, America’s yearly health care costs are about 17 percent of GDP,” Pattem said. “It is also well known that there are many areas of health care where money is being spent inefficiently.”

Saving money

According to the study, several opportunities exist for saving money without compromising health care quality. One of them is to prevent over-prescription of expensive brand-name drugs when equivalent and cheaper generics are available. According to Pattem’s research, these tendencies vary by state and drug class.

In addition, just as in every other field, peer influences are very powerful, and they play a role in medical prescription practices. Physicians very likely prescribe the same drugs as their colleagues. Pattem recommends a nuanced approach to physician awareness and policymaking that can take advantage of these findings.

“If we showed physicians in New Jersey and Hawaii that, as a group, they prescribe brand-name central nervous system agents at a rate 2.5 times greater than in Washington and Oregon, they would have to think about it differently,” Pattem said. “If physicians in West Virginia knew that they have the highest rates of brand-name prescriptions for metabolic drugs, but the lowest for psychotherapeutic drugs, how would they respond?”

Finally, companies market their branded versions of drugs as being new and different. These products are usually more expensive than their generic equivalents, even when they are just as effective. This has a direct effect on what is known as “patient-driven demand.” When people are exposed to drug ads that refer to the condition they have, they are influenced by the company’s marketing efforts and will be more likely to ask their doctors to prescribe them that exact brand name.

People tend to believe that expensive versions of drugs are more effective than cheaper ones.

Sundeep Pattem

“Generally, people tend to believe that expensive versions of drugs are more effective than cheaper ones,” Pattem said. “Marketing is really successful in getting that message across.”

A socially important cause

For this reason, Pattem believes that regulations on the kind of advertising permitted and the claims made about the effectiveness of drugs must be considered. In addition, the training that physicians receive should include ways to effectively deal with such demands from patients.

In September, Pattem presented his work at the Data for Good Exchange conference, a venue that connects data scientists with the public sector so that they can see the benefits of applied data science. Currently, he is a fellow with the California Department of Justice, working on the Open Justice platform and other innovative uses of big data to improve law enforcement.

“I am really delighted to see Sundeep apply his PhD training to do data analytics for a socially important cause,” said Bhaskar Krishnamachari, a USC Viterbi associate professor who advised Pattem during his doctoral studies. “I hope policymakers take notice of his striking findings that indicate that significant cost savings are possible in public health care.”

More stories about: , , ,

Can big data lead to lower costs for health care?

Top stories on USC News