Lessons Learned
The researchers stressed that data used in this type of decision-making needs to be publicly available for all counties. Using data that is not captured in every county could mean that the needs of some stakeholders are left underestimated. Similarly, some proprietary or private data — like Medicare data — may contain useful information but could take too much time to acquire and reduce transparency and reproducibility. For these reasons, the researchers recommend that decision-makers in similar situations use publicly available data that cover all affected people and communities.
Any formula used for distributing funds had to be sensitive to stakeholder opinions and needs, according to the researchers. Their first formula relied heavily on overdose-death statistics. That formula was less favorable to some counties because it did not account for costs related to overdoses where the patient may not have died. In response, the researchers incorporated data on the amount of naloxone distributed in each county and the number of overdose-related hospitalizations.
Additionally, the researchers found that they had to consider how rural counties would fair in an allocation formula relative to more populous counties. For the smallest counties, the initial formula would have generated allocations of just a few hundred thousand dollars over the course of 18 years. Even for a small county, this amount of money would likely be insufficient to do any meaningful work to abate the opioid problem. For that reason, the final allocation formula ensured that every county received a minimum amount of money, ensuring that smaller rural counties receive sufficient funds to expand prevention and treatment in their communities.
Once the adjustments were made, all counties agreed to the proposed funding model.
“The situation is complicated, so our job was to be impartial and to develop as simple a model as possible using transparent data,” Rhubart explained. “This was necessary in order to promote buy-in to the model in a timely manner. I hope that other groups working on a model like this can learn from our experience.”
The interdisciplinary team
Dennis Scanlon, distinguished professor of health policy and administration and director of the Center for Health Care and Policy Research, had worked with Pennsylvania’s government before on issues of health policy. He helped construct the team of researchers that included the expertise needed to develop the model.
“This project served an important public purpose while relying on the expertise and impartiality of a team of researchers at Penn State, Pennsylvania's land-grant institution,” said Scanlon. “In service of this important mission, the team worked quickly and directly with the counties and municipalities, their legal counsels, and the Pennsylvania Attorney General's Office. Everyone on the team understood the importance of allocating this huge sum of money fairly, so that Pennsylvania can reverse the tide of the opioid epidemic.”
Members of the research team included Rhubart; Scanlon; Qiushi Chen, assistant professor of industrial and manufacturing engineering; Glenn Sterner, assistant professor of criminal justice at Penn State Abington; Rob Newton, graduate student of industrial and manufacturing engineering; and Bethany Shaw, assistant director of data accelerator compliance at Penn State’s Evidence-to-Impact Collaborative.