At Bucknell University, students are using Pennsylvania Health Care Containment Council (PHC4) data in the classroom for researching health policy and analyzing health care utilization. In classroom exercises, students create descriptive statistics and perform analyses to identify factors affecting inpatient hospitalizations and health outcomes.
“The primary benefit has been giving undergraduates access to real life datasets,” said Amy Wolaver, PhD, associate professor of economics and co-director of the Bucknell Institute for Public Policy. “They get to practice data management skills and learn about research ethics and analysis.”
To date, Bucknell University has purchased PHC4 standard statewide inpatient discharge datasets for 2009 and 2012. A single-year statewide inpatient discharge dataset contains 2 million records. All students who use PHC4 data undergo training in data confidentiality.
A fruitful partnership between Bucknell University and PHC4 began in 2010, when Wolaver contacted the agency to find out how students might be able to support PHC4’s work while gaining experience in research and analysis. PHC4 and Bucknell University representatives met and decided the focus of the analyses should be hospitalizations for diabetes, a chronic condition that is the nation’s seventh leading cause of death. Three student interns—Hillary Freed, Mark Ayoub and Tom Shull—were selected to conduct a series of complicated data analyses on diabetes-related hospitalizations in Pennsylvania in 2009 using advanced statistical tools. PHC4 reviewed and verified the students’ analyses and merged them with data covering the period from 2000 through 2009. PHC4 released the resultant report, Diabetes Hospitalization Report, in November 2011.
“The Bucknell University students and their faculty supervisor who assisted in this research are to be commended for their excellent work on this report,” commented Joe Martin, PHC4’s executive director. Martin said the report provided important information on the long-term trends and impact of diabetes on the health care system in Pennsylvania.
Another ongoing project at Bucknell University is the building of a Geographic Information System (GIS) database comprised of a variety of data with which to explore health care utilization. The long-term goal is to link other geographic health measures—such as county health rankings on obesity and smoking rates from the Robert Wood Johnson Foundation—with measures of socioeconomic status. Students recently created GIS maps using aggregated PHC4 data, paired with census and other data, to evaluate how the supply of medical inventory in an area relates to the number of hospital admissions. Students hypothesized that where there are more staffed hospital beds per capita, there would be more admissions.
“Their hypothesis seemed to hold,” said Wolaver.
During the summer of 2013, two students, Michael McGown and Noelle Watters, set up a classroom exercise for the fall semester. The activity involved analysis of hospital admissions per 10,000 residents, adjusted for county age/gender distribution, for bacterial pneumonia, chronic obstructive pulmonary disease (COPD), knee replacement, gastroenteritis and prostatectomy.
PHC4 data may also be used in a study, to be conducted through a partnership with Bucknell University and Geisinger Health System, to explore the non-environmental aspects of Marcellus Shale drilling. (Geisinger Health System is conducting a well-publicized pilot study focusing on environmental impacts of drilling techniques on ground water and ambient air pollution.) Prior research has found some evidence of “social disruption” during previous boom-bust cycles related to resource extraction. The joint study would examine the socioeconomic and non-pollutant effects of drilling, such as motor vehicle accidents, treatment for alcoholism, and new diagnoses of major depressive disorder. A data validation study will determine whether medical claims data (PHC4 hospitalization data and Geisinger’s clinical data) is adequate to establish whether there are links between these negative health outcomes and Marcellus Shale drilling. If so, the partners will pursue grants to study the differences in non-environmental outcomes. The methods would compare drilling areas to non-drilling areas as a control group.
“PHC4 has great quality data, and the possibilities for analysis are endless,” Amy Wolaver said. “The hard part is narrowing down what we want to do with it.”