KIPRC researchers look into new data sources for rapid overdose surveillance

Dec 8, 2021

As public health surveillance works to increase the timeliness of data, specifically to identify changes in trends and to inform intervention strategies much closer to real time, researchers from the Kentucky Injury Prevention and Research Center (KIPRC) and the University of Kentucky looked at two new data sources for rapid drug overdose surveillance — emergency medical services (EMS) and syndromic surveillance.

In their article, “Emergency Medical Services and Syndromic Surveillance: A Comparison with Traditional Surveillance and Effects on Timeliness”, published in Public Health Reports, researchers compared the timeliness of those two data sources to the traditional use of emergency department (ED) billing data for nonfatal opioid overdose. The researchers concluded that data from EMS and syndromic surveillance systems can be reliably used to monitor nonfatal opioid overdose trends in Kentucky making data results available within days instead of months.

“Kentucky’s discharge data availability is one of the quickest in the nation, yet it is still only available in quarterly ‘chunks’ around 90 days after the end of a quarter—while national estimates are delayed up to three years,” said one of the report’s authors, Peter Rock, MPH. “When each quarterly chunk of Kentucky’s ED billing data is available, it is for occurrences that are three to six months old. Syndromic surveillance and EMS data are far more timely, with opioid overdose case data results being available for analysis within 24–48 hours and 60 hours, respectively.”

Rock, co-investigator on KIPRC’s CDC-funded Overdose Data to Action (OD2A) program, said the report compared these two novel rapid data systems to the traditional standard to evaluate their reliability for trend analysis.

Researchers were able to examine Kentucky’s syndromic surveillance (a rapid form of emergency department data) and the Kentucky State Ambulatory Reporting System (KStARS) emergency medical services run data in comparison with Kentucky's emergency department administrative claims billing data (a traditional morbidity surveillance data set). All three data sets use differing definitions for nonfatal opioid overdose based on the context of the data, Rock noted, but the researchers expected it would have meaningfully similar trends for public health surveillance purposes. Times-series analysis methods were utilized to compare how well the rapid systems and traditional data systems trend together.

According to Rock, the results of the trends, after adjusting for influences inherent to times series (e.g., autocorrelation), indicated that these time series follow similar patterns.

“When we see an increase in syndromic or EMS opioid overdose data, we would expect a similar increase in traditional ED data, when it becomes available after a significant lag,” he said.

An example of the lag reduction, which was included in the report, is the emerging trend of opioid overdose during the 2020 COVID-19 pandemic.

Rock said that a large portion of the spike was seen in 2020 second quarter (April–June) data, but this respective data from traditional sources would not be available for analysis until November 2020. However, the trend was realized almost immediately using syndromic and EMS data sources.

“Both data systems are far more timely than the traditional data system—and available in a few days as compared to months,” Rock said. “As mentioned previously, the traditional data are available in quarterly chunks 90 days after the end of a quarter. So on initial receipt that data is three to six months old (the end and beginning of that quarter); but, as real time progresses, the traditional data remain static until the next quarter closes, meaning that the lag just prior to the subsequent data receipt grows to six to nine months old.”

In addition to the direct findings of the paper in regard to EMS/syndromic surveillance compared to traditional standards, the report also presents a possible methodological framework for comparing a host of public health surveillance definitions in rapid systems versus traditional standards. The gold standard of a definition validation study is often rooted in medical chart reviews of identified cases, Rock said; however, these can be extremely resource-demanding. The approach presented in the report, he added, demonstrates a method for comparing novel definitions in rapid systems to more established definitions in traditional systems from a times-series trend perspective.

The article was co-authored by Rock, Dana Quesinberry, JD, DrPH, Michael D. Singleton, PhD, and Svetla Slavova, PhD. To read the report, visit https://journals.sagepub.com/doi/10.1177/00333549211018673.

Findings from the article will be presented by Rock to a national audience as part of an ASPPH (Association of Schools & Programs of Public Health) webinar on Dec. 7.

“The finding that the timelier data is in concordance with traditional surveillance data sources is important to share with our colleagues in the overdose surveillance community,” said Quesinberry, the principal investigator for surveillance on OD2A for KIPRC.

For more information on the webinar, visit https://www.aspph.org/event/public-health-reports-meet-the-authors-webinar-the-surveillance-of-nonfatal-and-fatal-drug-overdoses/.

KIPRC is a unique partnership between the Kentucky Department for Public Health (KDPH) and the University of Kentucky’s College of Public Health. KIPRC serves both as an academic injury prevention research center and as a bona fide agent of KDPH for statewide injury prevention.