Today, Sahan Journal published the first piece in a series investigating racial disparities in Minnesota’s opioid overdoses.
Sahan Journal’s health reporter, Sheila Mulrooney Eldred, has been covering health disparities for several years. When she started writing about the opioid epidemic in early 2023, she met a community leader named Abdirahman Mukhtar. Abdirahman, who spends every Friday night handing out pizza and Somali tea to community members affected by opioids, explained how helpful it would be to get accurate, detailed data on the epidemic.
Last fall, Sheila learned about a fellowship program through the Association of Health Care Journalists that would enable her to follow that question. So she applied.
During her reporting, she discovered that publicly available data, like the Minnesota Department of Health’s statistics on opioid overdose, provide only limited insights into racial disparities. These reports include broader racial categories, like Black or Asian, instead of breaking the numbers down by detailed ethnicities. From these aggregated statistics, it is unclear how the opioid epidemic impacts Minnesota’s Somali and Hmong communities, for example.
So a few key questions arose: How could we gain insights into the epidemic’s deadly toll for Minnesotans on a community level? And why are these datasets missing from public view?
For answers, we turned to a source of mortality data that is frequently studied by epidemiologists: death certificates.
Where did we get this data?
Fortunately, Minnesota is one of the only states in the U.S. where the full death certificate database is public. For this series, Sahan Journal requested and analyzed over 240,000 death certificates issued in Minnesota over the past five years to uncover hidden trends and disparities.
This data was obtained via a data request made by Sahan Journal to the Office of Vital Records at the Minnesota Department of Health. We requested all death certificates issued in Minnesota in the past five years. The office sent data in electronic form for every death certificate issued from 2019 to 2023. Each entry includes about 300 fields, including the decedent’s age, detailed ethnicity, causes of death, occupation, education level and more.
How did we find opioid-related deaths in this data?
In this section, I’m offering a detailed explanation of our extraction method for other data journalists and people who work in public health. General readers may want to skip to the next section — although the recording system for death records is actually pretty interesting.
I used ICD-10 codes to identify every record of opioid-involved overdose death from the death certificates we’d acquired. ICD-10 codes are a medical coding system that physicians use to classify and notate diagnoses and symptoms. I referred to the Minnesota Department of Health’s drug overdose death definition and methodology to extract opioid-related deaths from this data.
First, I extracted all records of drug overdose deaths, which include opioid and non-opioid overdoses. A drug overdose death is defined as having an underlying cause of death within the following range of ICD-10 codes:
- X40-X44: Accidental poisoning by drugs
- X60-X64: Intentional self-poisoning by drugs
- X85: Assault by drug poisoning
- Y10-Y14: Drug poisoning of undetermined intent
I then filtered the records to include only opioid-involved overdose cases. An opioid-involved overdose is defined as having a contributory cause of death within the following ranges of ICD-10 codes:
- T40.0: Poisoning by Opium
- T40.1: Poisoning by Heroin
- T40.2: Poisoning by Other Opioids
- T40.3: Poisoning by Methadone
- T40.4: Poisoning by Other Synthetic Narcotics
- T40.6: Poisoning by Other and Unspecified Narcotics
These drug categories are not mutually exclusive. For example, a death involving heroin (T40.1) and methadone (T40.3) will be counted in both drug categories.
Using this extraction method, I identified a total of 4,116 opioid-involved overdose deaths that occurred in Minnesota from 2019 to 2023.
SAHAN INVESTIGATES: RACIAL DISPARITIES AND FENTANYL DEATHS
Overlooked: Who suffers the most from the opioid epidemic in Minnesota?
Opioids have killed more than 4,000 people in Minnesota over the past five years. Who are these Minnesotans? Sahan Journal looked at death records to find out which communities are most at risk. We confirmed that the problem is far worse for people of color in Minnesota. And, with new data and greater clarity, we’re showing how the disparities are impacting specific communities: From 2019 to 2023, Native Americans were at least 15 times more likely to die from opioid overdoses than white people. Somali Minnesotans were at least twice as likely to die from opioid overdose than their white…
Keep readingHow did we calculate opioid overdose death rates for different communities?
People of color make up only 23% of Minnesota’s population. For communities with smaller populations, the death counts tend to be smaller than those of the white community.
To make a fair comparison between opioid overdoses among communities of color and white Minnesotans, I converted absolute numbers of opioid overdose deaths into crude rates. A crude rate is calculated by dividing the number of deaths in a particular population by the total number of people in that population and then multiplying that ratio by 100,000.
Crude Rate = Count / Population * 100,000
For example, 259 Black Minnesotans died of opioid overdose in 2023. The death rate is calculated by dividing 259 by the total Black population in Minnesota (398,326). Finally, we multiplied that number by 100,000. Following this formula, we got a death rate of 65 per 100,000 Black Minnesotans.
For our analysis, the population of different subgroups is sourced from the 2022 American Community Survey (5-year estimates).
As we dived deeper into the data, we started to understand why data for smaller ethnic groups is hard to scope out.
When the actual numbers of deaths are small, the mortality rates calculated with these numbers tend to be unreliable. The U.S. Centers for Disease Control and Prevention (CDC) marks mortality rates as “unreliable” when the death count is less than 20. In addition, all statistics representing fewer than 10 persons are suppressed to protect the privacy of decedents.
Misclassification is another obstacle in calculating detailed data for smaller populations. Errors tend to happen when categorizing people by ethnic identity. During our reporting, we interviewed health experts and community leaders to get a deeper understanding of this challenge. We also documented the efforts that different groups have made to obtain better data on opioid overdose in their communities.
Data doesn’t always tell the truth. In fact, data is a continuation of biases and errors that humans hold. In our article, we turned to data to gain insights into trends of opioid overdoses in Minnesota. Still, the story of the opioid epidemic’s impact belongs to people who are experiencing despair first-hand, not to the numbers.
In upcoming stories, then, we will feature more community voices to learn the impact of the opioid epidemic on smaller racial groups. We will also share more data to highlight trends — and possible solutions — within these communities.
If you have a story to share, contact Sheila Eldred at seldred@sahanjournal.com.
Have questions about this data and our analysis? Contact Sahan Journal’s data reporter, Cynthia Tu, at ctu@sahanjournal.com.
The series is part of a reporting fellowship sponsored by the Association of Health Care Journalists and supported by The Commonwealth Fund.
