Hidden Fentanyl Is Driving a Fatal New Phase in US Opioid Epidemic

Entry type: Single project

Country/area: United States

Publishing organisation: Bloomberg News

Organisation size: Big

Publication date: 2022-12-13

Language: English

Authors: Reporting by Emma Court
Data and analysis by Linly Lin and Madeline Campbell
Graphics by Leonardo Nicoletti
With assistance from Jeremy Diamond and Paul Murray
Edited by Tim Annett, Chloe Whiteaker, David Ingold and Drew Armstrong


Emma Court is a health reporter in New York, Linly Lin is a data journalist in San Francisco, Madeline Campbell is a data journalist in New York, Leonardo Nicoletti is a data visualization journalist in Rome.

Project description:

This data-driven investigation sheds light on the devastating consequences that the rise of the synthetic opioid fentanyl has caused for US society. Since 2000, more than half a million Americans have died from opioid overdose. The majority of these deaths have occurred relatively recently, as the Covid-19 pandemic contributed to circumstances that led fentanyl use to spiral out of control. Through an analysis of 20 years of CDC data and interviews with local officials and advocates in Volusia, Florida, this article illustrates how the pandemic and fentanyl were twin forces that transformed the opioid epidemic into a nationwide fentanyl crisis.

Impact reached:

Extensive reporting, data analysis and visual storytelling shed light into the rise of fentanyl in the context of the Covid-19 pandemic.

Shortly after publication, Florida senator Rick Scott held a roundtable discussion about fentanyl in Volusia County that included representatives from the Volusia Recovery Alliance as well as the sheriff’s office who were highlighted in our story. The Volusia sheriff’s department was also featured in a TV segment on NewsNation to talk about the issue because of our story.

In addition to illustrating the temporal and geographical distribution of deaths caused by fentanyl, the story highlighted new alarming trends such as the recent demographic shifts in who opioids kill. Historically, opioid overdoses were killing mostly White Americans, but now Black Americans are dying from opioids at much higher rates than other races. Death rates among other minority groups also grew disproportionately during the pandemic.

Techniques/technologies used:

We did an extensive study into the CDC’s epidemiologic research data portal, CDC WONDER, and the International Classification of Diseases (ICD) standards to understand how causes of death are identified, categorized and reported. Using Python, we sliced opioid overdose-involved deaths by demographic and geographic variables and landed on a precise ICD code combination, spelled out in our methodology section, to analyze opioid overdose death trends and the relative lethality of the drugs within this class.

The entire reporting process was guided by the findings from our data analysis. Volusia County, Florida, a semi-urban coastal region that wouldn’t be a traditional place of choice for a drug overdose story, was a standout when we compared changes in US counties’ fentanyl death rates.

We combined cutting-edge technologies and data visualization techniques to develop a series of engaging interactive graphics using Svelte and D3.js.

An important objective was to prompt the reader with a strong sense of urgency about the human and societal tolls of fentanyl. We sought to accurately represent the scale and somber nature of the problem, while at the same time humanizing each data-point. For example, the stacked time-series of “Deepening Crisis of Opioid Deaths in the US” is meant to evoke a sense of mortality – the rows of reddish-hued dots drip down the page like blood dripping from a wound. In stacking thousands of circles on top of one another and allowing the graphic to overflow outside of the screen’s lower boundary, the graphic strives to represent the actual people that died of fentanyl overdose in a more tangible way than a bar or line could accomplish. This task required the visualizing large amounts of data in the browser by leveraging technologies not traditionally used for data-visualization, such as the use of html canvas elements to render graphics.

Context about the project:

Throughout the reporting process, we ran into many issues involving data accuracy, which is an inherent challenge when journalists work on illicit drug-related subjects, but this particular one showed that there’s a lack of accountability in the government system in tackling drug overdose as a national crisis.

The data request for hospitalization data placed to Florida’s governmental health agency went on for more than three months. We were redirected a couple times but remained in active correspondence for weeks. However, the first versions of data provided by the government were completely wrong. We had to persistently follow up and guide the agency to pull accurate data. During the process, we also learned that fentanyl wasn’t tracked as a cause for emergency room visits in Florida until after 2020. This discovery confirmed our assumption that fentanyl’s impact had been overlooked at the local level and policy makers were reacting with a delay.

The Substance Abuse and Mental Health Services Administration, an agency within the US Department of Health & Human Services, is tasked with surveying and tracking the number of people using opioids in the US and reports annual data. Yet, when we were interviewing the agency, a spokesman informed us that their publicly disclosed data for recent years was inaccurate and shouldn’t be used for analysis. This makes us question the agency’s effort to carry out its mandate. We also pressed the administration to provide spending data on overdose prevention, but such data were said to be non-existent – making it apparent that there is much to be done to improve accountability around efforts to reverse the trend.

What can other journalists learn from this project?

Often the data that makes it into the final version of a story is only the tip of the iceberg of data that was collected, considered, and sifted through in order to contextualize the landscape surrounding an issue. The CDC WONDER data used extensively in the story proved the most complete and reliable data source available, though countless versions of the data with temporal, geographic, demographic and ICD-subset splits were considered in the process of understanding the opioid epidemic from a data perspective.

Many other pieces of data – federal, state and local – were examined in an effort to piece together the social, economic and law enforcement consequences of the opioid epidemic in Florida and beyond. Some of this data was readily available but most of it required elaborate data requests and correspondence.

In some cases we learned there was no data available to answer questions we’d hoped to consider. Both policy makers and civilian advocates are operating without a full understanding of this deadly epidemic’s landscape. Sometimes there are insights to be gleaned from what data isn’t available as compared to what is reliably collected.

We came to realize that persistently asking and pursuing data-related questions is how journalists could help the government and industry players to better improve their data efforts. Data issues at government agencies may surface; sometimes there is an underlying issue but other times it may not be intentional. We learned to expect that mistakes in data collection and reporting could happen, and many problems and errors could be resolved through communication and education. We are also contributing by pointing out the mistakes and helping the government further improve data accuracy and transparency.

Project links: