Association for Public Policy Analysis & Management

The role of prescription opioids in the current opioid crisis has been well-established, with much of the research focus on their relationship with overdose deaths. For example, the CDC reports that prescription opioid-related death rates rose from around 3 per 100,000 in 2008 to 4.5 per 100,000 in 2016. How the availability of prescription opioids in a community relates to opioid-related hospitalizations has been less studied. Hospitalizations are a distinct indicator of substance use, have cost implications for health care systems, and may also be an entry point into substance use treatment as well as other support services. Using an instrumental variables approach on county-level data for 2011 through 2014 for most counties in the U.S., we show how changes in retail opioid sales predict opioid-related hospitalization rates.

Data and Methods

Our data cover most counties in the U.S. from 2011 through 2014. Data on opioid-related hospital stays are from voluntary reports from states on patient-level hospital stays, to the Healthcare Cost and Utilization Project in the Department of Health and Human Services. We measure stays as rates per 100,000 people. Data for retail prescription opioid sales comes from the Drug Enforcement Administration’s Automation of Reports and Consolidated Orders System (ARCOS). ARCOS reports contain information on the inventories, acquisitions, and dispositions of certain controlled pharmaceuticals. We selected commonly prescribed and misused opioids, and which were consistently reported to ARCOS over the time period of study.

Being an administrative data collection with spatial and temporal data reporting issues, ARCOS data likely contain substantial measurement error. I use an instrumental variables approach to account for this error, using Medicare Part D opioid prescription rates as an instrument. Two-stage least squares models were run, with county and year fixed effects using robust standard errors. Additional control variables include county demographics and economic measures, uninsurance rates, physicians per capita, and presence of a prescription drug monitoring program.

Results and Discussion

Nationally, increases in rates of retail opioid sales predict lower rates of opioid-related hospitalizations. Urban and rural counties present starkly different relationships between retail sales rates and opioid-related hospitalization rates. In urban areas, a 10 percent increase in sales rates predicts a 1.7 percent decline in opioid-related hospitalizations. Rural areas show the opposite effect – a 10 percent increase in sales rates predicts a 2.3 percent increase in opioid-related hospitalizations.

The difference between urban and rural counties suggests different mechanisms link prescription opioids to hospitalizations. For example, it could be that the availability of illicit opioids in urban and rural areas accounts for some of the differences. If urban areas have a greater prevalence of illicit opioids, a decline in prescription drugs leads users to these more dangerous drugs, leading to higher hospitalization rates. On the contrary, in rural areas, prescription opioids may be what drives hospitalization rates if illicit opioids are not as widely available.