The University of Pennsylvania’s Wharton School and Leonard Davis Institute of Health Economics are partnering with health care insurance giant Anthem, Inc. in a research collaboration offering academic researchers unique new sorts of health services research data.

Headquartered in Indianapolis, Anthem is the country’s second largest health insurer with $80 billion in annual sales and 44 million customers. Anthem’s part of the collaboration will be handled by its Wilmington, Delaware-based subsidiary HealthCore, a clinical outcomes and health data firm with access to its parent company’s medical claims, pharmacy claims and laboratory results.

Anthem’s HSR team
HealthCore is also Anthem’s in-house health services research center that employs a team of 15 scientists including specialists in economics, epidemiology, industrial engineering and decision science.

An executive steering committee comprised of representatives from all three organizations will consider and greenlight research project proposals.

The collaborative program anticipates conducting studies in potential areas including health care informatics and economics, incentives for both providers and patients, organizational models, benefit design, pharmaceutical formulary practices, bundled payments, medication adherence and alternative payment systems.

LDI the collaborative hub
“Here at LDI, we bring together investigators from throughout the university to collaborate on innovative projects and investigations,” said LDI Executive Director Daniel Polsky, PhD. “Our vision at LDI is affordable, accessible health care for everyone. It’s clear from our work with Anthem that they’re interested in finding solutions to those same challenges. What’s exciting about this is the willingness of Anthem to give us the academic independence to really explore the twists and turns and really discover what the data are telling us.”

Guy David, the Lead Coordinator of the initiative, emphasized that it is not a transactional arrangement — Penn is not purchasing data from Anthem nor selling analytical services. Instead, it is a genuine collaboration among scientists at both HealthCore and Penn to advance knowledge about how to improve the nation’s health care.

“This is a really rare opportunity for the Penn research community,” said David, PhD, an Associate Professor of Health Care Management at the Wharton School and an LDI Senior Fellow. “We’re talking about claims data for 44 million people spread across commercial, Medicare Advantage and Medicaid populations. The sheer scale of this data and its span across so many socioeconomic levels and states represents a remarkable resource of the sort we haven’t had before.”

‘Exploring outside ideas’
David’s counterpart at HealthCore, Andrea DeVries, PhD, emphasized the mutual benefit for the collaborators as well as science itself. “There’s so much to know when it comes to health economics and health services research,” she said. “We appreciate the opportunity to connect with other researchers who have specialized expertise that we might not have. I think we all have the goal of exploring outside ideas and learning something new.”

Devries, who is HealthCore’s Senior Director of Payer and Provider Research, also noted the benefit of having the opportunity to interact with future researchers in Penn’s graduate programs.

“Anthem can bring a lot to Penn’s educational mission,” said David. “It will be a game changer for our doctoral students to write their dissertations using this very clean and precise data resource that can make their work more impactful,” he said.

Access to data managers
Penn faculty researchers point out that one of the most alluring aspects of the new collaboration is direct access to the people who manage Anthem claims data — an issue whose critical importance is not well understood outside the research community.

Associate Professor at Penn’s Perelman School of Medicine, Peter Groeneveld, MD, noted that although a growing number of insurers have recently begun sharing commercial claims data with outside researchers, those data sets are often peppered with fields of enigmatic content. For instance, health plan data from Insurer A based in Florida may have different data and coding standards than another Insurer A office in California. When those two streams are mashed together into a “national” data set for distribution to researchers, there are inevitably data fields or variables that make no sense.

Groeneveld, a former computer engineer who directs the LDI Health Services Research Center, said academic investigators who acquire and use such data sets are rarely able to communicate with the technical professionals who assemble them — a barrier that often makes it impossible to learn anything more about what a specific data field or variable means.

‘Great deal of difference’
“The idea that this new collaboration would connect Penn people to the Anthem people who can be the translators could make a great deal of difference,” Groeneveld said.

Penn health economist Mark Pauly, a member of the collaboration steering committee, agrees. “There’s often a lot of weird stuff that comes out the first time they run the data, like fields full of zeros. And usually, there’s a reason for it at the same time there’s no one to ask about that reason. So, you’re sitting there alone, asking yourself ‘why does my data look so crazy?’. But the people who actually ran the data don’t want you bothering them,” he said.

Pauly, PhD, is a Wharton School Health Care Management professor and LDI Senior Fellow who has 50 years of experience in health services research.

He also pointed out that another unusual aspect of the collaboration is Anthem’s willingness to evaluate innovative, evidence-driven ideas using real-time pilot studies. Essentially, Pauly said, this would enable researchers to quickly test a hypothesis to see if “there’s a ‘there’ there” that justifies writing a longer proposal for a larger research effort.

“Overall, this new relationship is really exciting,” said David. “I’ve long lived in this frustrating world where research is carried out with unstructured data you scraped or with data you pay enormous amounts of money for and in many occasions turns out to be insufficient. I kept telling myself there’s GOT to be a way to get actual data and not crumbs from institutions who generate data as a byproduct of operations and not for the purpose of advancing knowledge. Now, with Anthem and its vast pools of claims data, we’re working with insurance decision makers who want a partnership that will help them extract new knowledge from their data. It feels to me like a real win-win situation.”