Examining trends in polydrug overdose deaths across rural Midwest counties in the United States, 2022 through 2024

Abstract

Purpose

To examine trends in overdose deaths and the presence of xylazine in postmortem toxicology in a highly rural region of northern Minnesota with a significant American Indian/Alaska Native (AI/AN) population, using real-time local surveillance data.

Methods

We analyzed drug overdose death data from the Midwest Medical Examiner’s Office, covering 36 counties from January 1, 2022 to December 31, 2024. Data included demographic characteristics, place of death, and substances detected in toxicology results. Age-adjusted mortality rates and disparities across racial/ethnic groups were calculated, and polydrug combinations were explored using network analysis.

Findings

Among 967 overdose deaths, most decedents were white (71.7 %), AI/AN (13.4 %), or Black/African American (10.3 %) with an overall mortality rate of 76.1. Mortality rates declined during the study period for white and AI/AN populations and increased slightly for Black/African American populations. However, rates remained disproportionately high for AI/AN populations, who were 6.19 times more likely to die of overdose than non-AI/AN in 2024. Xylazine was detected only in combination with fentanyl, with no racial/ethnic differences in its presence. Distinct polydrug patterns were observed by race: fentanyl-cocaine combinations were more prevalent among Black/African Americans, while fentanyl-methamphetamine combinations predominated among AI/AN and White decedents.

Conclusions

Local surveillance in rural areas can detect emerging threats like xylazine and illuminate racial disparities obscured in national datasets. Findings highlight the urgent need for real-time, locally driven data to inform targeted prevention and harm reduction, particularly for AI/AN populations in rural communities.

Introduction

The United States Centers for Disease Control and Prevention’s (CDC) National Vital Statistics System (NVSS) reported more than 100,000 accidental drug overdose deaths in 2022 with more than one million deaths in the past two decades, underscoring an ongoing and rapidly evolving public health crisis (CDC, 2021; CDC, 2024). A majority of overdose deaths, in both rural and urban areas of the United States, are attributed to fentanyl, an illicitly manufactured synthetic opioid with a growing myriad of analogs (i.e., chemical variants) (Armenian et al., 2018; Park et al., 2021). Fentanyl first appeared in the Northeast around 2012, then proliferated across the Midwest around 2014, and finally dominated the West Coast market by late 2019 (Hedegaard et al., 2019; Kolak et al., 2020; Lopez et al., 2024; Peters et al., 2020; Ruhm, 2017; Victor et al., 2023). Notably, the early identification of fentanyl in these regions were not provided by national surveillance data, but rather from local surveillance systems that examined toxicology results and death certificates from death investigation agencies (Algren et al., 2013; Krinsky et al., 2011; Martin et al., 2006; Ogilvie et al., 2013; Okic et al., 2013). Before 2013, fentanyl was grouped under broader drug categories in the NVSS, leading to frequent misclassification and underreporting (Peppin & Coleman, 2021; Ruhm, 2016; Seth et al., 2023). It was not until 2013 that fentanyl was categorized separately in national mortality data, enabling more accurate tracking of its role in overdose deaths.

Local surveillance systems have proved critical for identifying emerging adulterants in the North and South American drug supply, such as xylazine, a veterinary sedative not approved for use by humans that has increasingly been detected in the illicit fentanyl supply (Alexander et al., 2022; Meyer et al., 2023). Xylazine use has been linked to negative health consequences, including more severe and altered overdose and withdrawal symptoms, as well as skin ulcers and tissue wounds, some of which have required amputation (Quijano et al., 2023; Zagorski et al., 2023). Like fentanyl, xylazine has primarily been tracked through local surveillance systems and monitoring efforts have largely been limited to urban areas (Johnson et al., 2021; Kariisa, 2023; Nunez et al., 2021; Reyes et al., 2012; M. Spencer et al., 2023; Wong et al., 2008). Moreover, national mortality data systems rely on International Classification of Diseases (ICD) codes for cause-of-death reporting, but there is currently no specific ICD code for xylazine limiting national tracking. Limitations with these data are further compounded by significant delays in data availability as national mortality data are typically released 12 to 18 months are the end of the calendar year (CDC, 2025; Slavova et al., 2019). For example, while provisional data suggest a potential decline in overdose deaths in 2023, official statistics will not be released until well into 2025. Similarly, data for 2024 will not become available until 2026.

As trends in illicit drug markets have shifted, so too have racial demographics of the overdose population. Prior to widespread emergence of fentanyl, overdose deaths were predominantly associated with as white rural populations; however, over the past decade, overdose mortality have increasingly affected Black and African American communities along with American Indian and Alaska Native (AI/AN) populations (Bauer et al., 2024; Friedman & Shover, 2023; Furr-Holden et al., 2021; Hedegaard & Spencer, 2021; A. Jones et al., 2023; Phalen et al., 2018; Smith et al., 2024; Townsend et al., 2022). A recent analysis of CDC data from2019 to 2022 found that opioid-related overdose deaths increased most rapidly in Black, AI/AN, and Hispanic/Latino Americans, with Minnesota exhibiting the highest rates of overdose deaths among AI/AN individuals (Smith et al., 2024). The authors note that to conduct this analysis, the data were pooled across 5 years due to guidelines related to data-suppression. Due to concerns of re-identification, county-level overdose counts below 10 are suppressed (Tiwari et al., 2014). While this policy protects individual privacy, it poses challenges for analyzing overdose trends among racial and ethnic minority groups in less populated areas, such as rural communities (Stone, 2025).

In this study, we analyzed near real-time mortality data collected from multiple US counties in northern Minnesota, an area characterized by extremely rurality, as part of overdose surveillance efforts funded from opioid remediation initiatives. These counties have a high proportion of AI/AN residents, so we examined population-adjusted overdose trends across racial and ethnic groups from January 1, 2022, to December 31, 2024. Additionally, we conducted a polydrug network analysis across these groups using toxicology results to explore patterns of substance co-use and assess the prevalence of xylazine use.

Data and Methods

Information on overdose deaths in this study comes from a partnership in Minnesota between the Midwest Medical Examiner’s Office (MMEO) and the Anoka County Public Health and Environmental Services. This collaboration aims to improve overdose surveillance efforts and inform targeted overdose prevention efforts (Anoka County, 2025; CDC Foundation, 2024). The MMEO is the largest medical examiner’s office for the state and provides death investigation services for 36 counties, covering approximately 1.8 million people and half of Minnesota’s geographic area. Three-quarters of the counties (75%; n=27) are considered completely rural, coded as non-metropolitan and rural, with an average Rural-Urban Continuum Code of 7.5. The RUCC ranges from 1 (most urban) to 9 (most rural), with higher values indicating greater rurality (Figure 1) (Butler, 1990; Rennison & Mondragon, 2022). The catchment area includes the Iron Range, a geographically and historically distinct rural region characterized by low population density, dispersed settlements, and a history of iron ore mining that has been shaped by cycles of industrial development and decline (Homans & Ackerman, 2025; Nelson, 2025; Whitson, 2019).

Figure 1.

Minnesota catchment counties where the Midwest Medical Examiner’s Office investigate overdose death, January 1, 2022 through December 31, 2024.

Notes: Rural-Urban Continuum Code (RUCC) from 1–9 with higher numbers indicating more rural areas. Darker gray and asterisks indicates catchment counties on the Iron Range. The catchment counties and corresponding RUCC include Cook (9), Grant (9), Lac Qui Parle (9), Lincoln (9), Swift (9), Pope (8), Chippewa (7), Renville (7), Yellow Medicine (7), Douglas (6), Itasca (6),* Koochiching (6), Lake (6),* McLeod (6), Meeker (6), Morrison (6), Otter Tail (6), Pine (6), Todd (6), Wadena (6), Kandiyohi (4), Beltrami (3), Rice (3), Stearns (3), Carlton (2),* St. Louis (2),* Anoka (1), Benton (1), Carver (1), Chisago (1), Isanti (1), Mille Lacs (1), Sherburne (1), Sibley (1), Wright (1).

The Anoka County Public Health and Environmental Services implemented a data collection system for extracting information from death certificates used in prior research (Lowder et al., 2018; Ray et al., 2017, 2020; Victor et al., 2023). The MMEO collects investigative information from deaths that are sudden, unexpected, unnatural, or suspicious to ensure accurate determination of cause and manner of death. This includes publicly available data on all suspected unintentional poisoning deaths, classified under ICD-10 codeX40-X44 (Sec. 144.225 MN Statutes, 2024). Throughout the data collection process, an abstractor coded information from investigation reports, with a senior reviewer conducting random checks for accuracy. The collected information included location of the fatality, decedent demographics (age, race/ethnicity, sex, and marital status), military status (active duty, veteran, reserves or non-military) and indicators for select substances detected in post-mortem toxicology screenings. In this analysis we focus on those implicated in rising overdose rates (particularly in rural areas) (Carpenedo Mun et al., 2023; Hedegaard & Spencer, 2021; Peters et al., 2020), opioids (fentanyl, heroin, and prescription opioids) and stimulants (cocaine, methamphetamine), as well as xylazine as an emerging drug in the illicit supply (Alexander et al., 2022; Quijano et al., 2023; Zagorski et al., 2023).

Analysis of these data were conducted as part of a broader analysis into the detection of xylazine in the illicit drug supply (R21DA060954) with this portion of the study not involving human subjects (RTI IRB# STUDY00022831). We begin with descriptive and bivariate statistics of the overdose events in the catchment area, followed by the calculation of age-adjusted overdose mortality rates (annual and quarterly) and rate ratios to compare prevalence odds ratios (Supplemental Table 1) by racial groups (United States Census Bureau, 2024) using Z-tests with a Bonferroni correction. Following this trend analysis, we examine polydrug combinations in toxicology findings using network analysis techniques for relational structures in text data (Carley, 1997; Ray et al., 2020). Statistical analyses were conducted in R version 4.4.2.

Results

Demographic patterns in overdose

Over the 3-year study period (January 1, 2022, to December 31, 2024), data were collected from 967 drug overdose deaths in 36 counties (Table 1). The average age was 42.8 years (standard deviation [SD]=14.0), and more than two-thirds of deaths were among males (70.0%). To avoid the risk of re-identification (counts that would indicate a single person) we use the following race-ethnicity categories in presenting our results, white (71.7%), AI/AN (13.4%), Black/African American (10.3%), and Other, Unknown and Multi-Race-Ethnicity (4.6%). More than two-thirds of the events took place in the persons residence (66.4%), 25.3% were recorded as having military status, and 61.5% of decedents were never married (Table 1).

Table 1.

Sociodemographic characteristics of overdose deaths in Minnesota catchment counties, January 1, 2022, to December 31, 2024.

	Overall		2022		2023		2024
Characteristics	967		377		336		254
Age	M	SD	M	SD	M	SD	M	SD
	42.81	13/98	41.36	13.61	43.00	13.96	44.70	14.38
Race	N	%	N	%	N	%	N	%
American Indian/Alaska Native	130	13.44	51	5.29	52	5.39	27	2.8
Black/African American	100	10.34	33	3.42	40	4.15	27	2.8
White	693	71.66	274	28.42	229	23.76	190	19.7
Other, Unknown and Multi-Race-Ethnicity	44	4.55	19	1.97	15	1.56	10	1.03
Sex
Male	664	68.7	264	70	231	68.8	169	66.5
Female	303	31.3	113	30	105	31.3	85	33.5
Marital Status
Married	128	13.2	50	13.3	46	13.7	32	12.6
Divorced	200	20.7	76	20.2	67	19.9	57	22.44
Never married	595	61.5	240	63.7	204	60.7	151	59.45
Widowed	17	1.8	4	1.1	5	1.5	8	3.2
Unknown	27	2.8	7	1.9	14	4.2	6	2.4
Place of Death
Residence	642	66.4	258	68.4	215	64	169	66.5
Hospital Setting	168	17.4	64	17	58	17.3	46	18.1
Other Residence/Unknown	157	16.2	55	14.6	63	18.8	39	15.4
Military
Non-Military	722	74.7	288	76.4	180	53.6	254	100.0
Military	245	25.3	89	23.6	156	46.4	0	0

Notes: Information based on death investigation records. As noted above, 71.66% of decedents were coded as white, this was followed by 13.44% AI/AN, 10.3%, Black/African American, 1.1% Asian and 1.0% Hispanic. Race categories above avoid re-identification when presenting analysis by year.

We examined patterns between race-ethnicity categories, substances detected in toxicology results, and social factors. Among substances detected in toxicology results, 85.3% contained opioids or stimulants; 61.2% were opioid-related, with 55.2% as fentanyl, 1.7% as heroin, and 11.7% as prescription or other opioids. We found that AI/AN (67.7% vs. 53.3%; χ2 = 8.87, Cramer’s V=.15, p < .01) and Black/African American (74.0% vs. 53.1%; χ2 = 15.07, Cramer’s V=.2, p < .01) decedents were significantly more likely to have fentanyl detected than other race-ethnicity categories. Additionally, AI/AN decedents were less likely to die at their residence (55.4% vs. 68.1%; χ2 = 8.59, Cramer’s V=.09, p=.01), and more likely to have never been married (75.4% vs. 59.4%; χ2 = 15.71, Cramer’s V=.13, p<.01).

Demographic trends in overdose over time

Overall, the age-adjusted overdose mortality rate among the catchment area was 76.1 per 100,000. This is higher than the state average of 24.8 from 2022, and higher than the national average of 32.6 for the same year (CDC, 2020; M. R. Spencer et al., 2024). In looking across the study period, overdose deaths showed a decrease that is consistent with provisional national trends from 2022 to 2023 (Kiang & Humphreys, 2025; Post et al., 2025). We found a reduction in overdose deaths, from 377in 2022 to 254 in 2024, and examined quarterly trends in age-adjusted overdose mortality rates by race-ethnicity categories (Figure 2). Overall, rates declined for white, AI/AN, and Black/African American populations, with decreases of a 34.7%, 46.5%, and 21.1% respectively. While AI/AN and Black/African American populations experienced a slight increase in 2023 (4.1% and 10.3%, respectively), rates declined in 2024. Rates among the White population declined steadily across all three years; however, year-to-year comparisons of rate ratios did not yield statistically significant changes for AI/AN or Black/African American populations. Over all years AI/AN individuals experienced higher mortality rates than white and Black/African American populations. In 2023, the overdose mortality rate among AI/AN individuals were 12.61 times higher than whites. Although rates decreased in 2024, the mortality rate among AI/AN people remained 8.04 times higher than white individuals. Furthermore, AI/AN individuals were still 6.2 times more likely to experience a fatal overdose as compared to non-AI/AN individuals (down from 8.6 in 2023 and 7.6 in 2022) (see Supplemental Table 1).

Figure 2.

Overdose mortality rates among American Indian/Alaska Native, Black /African American, and White Populations in Minnesota catchment counties, January 1, 2022 through December 31, 2024

Notes: Rates were age-adjusted using quarterly overdose mortality rates per 100,000 population from the US Census Bureau, with underlying quarterly rates detailed in Supplemental Table 1. The 2023 US Census Bureau population estimate was used for both 2023 and 2024.

Polydrug analysis

We used the toxicology results to conduct polydrug analysis. Most overdose deaths contained multiple substances; 57.2% of deaths with stimulants also included opioids and conversely, 52.7% of deaths with opioids also contained stimulants. Xylazine was detected in 4.1% of deaths overall (1.6% in 2022, 0.9% 2023, 1.8% 2024) and consistently detected alongside fentanyl; 22% of cases where xylazine was detected also involved cocaine, 56% involved methamphetamine, 20% involved alcohol, and 34% involved non-fentanyl opioids, including heroin or prescription opioids. In addition, there were no statistically significant differences in detection of xylazine by race-ethnicity.

Figure 3 provides a visual representation of polydrug combinations by race-ethnicity and year. For example, in 2022 there were 33 overdose deaths among Black/African Americans, 84.8% (n=28) of those contained fentanyl, and among those 32% also contained cocaine and 25% methamphetamine. In that same year there were 51 overdose deaths among AI/AN decedents, 70.1% (n=36) of which contained fentanyl, and among those 2.7% also contained cocaine and 47.2% methamphetamine. Among the 274 white decedents, 52.2% (n=143) contained fentanyl, and 9.7% of those also contained cocaine and 39% methamphetamine. Thus, as shown in Figure 3, without notable change over time, cocaine and fentanyl were consistently more common in the polydrug pattern among those coded as Black/African American, while methamphetamine and fentanyl were more common among AI/AN and white race-ethnicity categories. We also include additional visualizations of polydrug patterns without xylazine (Supplemental Figure 1) and an UpSet plot that shows counts of specific polydrug use patters by race (Supplemental Figure 2).

Figure 3.

Network of substances detected in overdose deaths by racial groups, January 1, 2022, to December 31, 2024.

Notes: The weight (thickness) of each line indicates the frequency (count) of co-occurrences between the connecting nodes (i.e. substances) in a given year by race-ethnicity. The node size indicates the frequency (count) of each specific substance with larger nodes denoting more frequent polydrug occurrences. Missing nodes indicate that a given drug was not detected in a year, while missing edges indicate that two drugs did not co-occur in that year.

Discussion

In this study we examined overdose data from northern rural Minnesota, which encompasses a uniquely rural region of the US (Nelson, 2025; Whitson, 2019). Consistent with provisional national counts from the CDC, we found a substantial decrease in overdose deaths between 2022 and 2024 (Ahmad et al., 2024). Despite these overall reductions, we found large disparities by race; for every overdose death among a white individual, there are six among AI/AN people, a disparity that highlights profound inequities in exposure, access to care, and structural determinants of health (D. Jones, 2006; Kruse et al., 2022). Despite reductions in overdose mortality for AI/AN and Black/African American populations, changes in age-adjusted rates were not statistically significant. Consistent with other research from the Midwest (Lopez et al., 2024; Smith et al., 2024; Victor et al., 2023), we found that in this study’s catchment area, where Black/African American populations are under-represented, overdose mortality rates among Black/African American individuals remained consistently higher than whites. Additionally, in this region where about 6% of the population are military veterans (United States Census Bureau, 2024), we found more than a quarter of overdose deaths were coded as having military status (Vakkalanka et al., 2025). In terms of toxicology results, fentanyl was detected in more than half of the overdose deaths and was more frequently detected among AI/AN and Black/African American individuals. There were discernable polydrug patterns by race-ethnicity categories; fentanyl was most often co-detected with methamphetamines among white and AI/AN individuals, and fentanyl was more commonly co-detected with cocaine among Black/African American individuals. These findings align with previous studies suggesting co-use patterns with stimulants have manifested since the rise of fentanyl (Friedman & Shover, 2023) and further suggests cultural preferences and drug distribution patterns (Hoebert et al., 2017; May & and Hough, 2004). Finally, xylazine, a relatively new additive to the illicit drug supply (Alexander et al., 2022; Meyer et al., 2023), was detected in 4.3% of overdose events, decreasing slightly in 2023 before increasing again in 2024, and was only detected along with fentanyl. There were no differences in the detection of xylazine by race-ethnicity categories (Supplemental Figure 1).

Some of these findings are highly novel, such as xylazine presence, because such data are currently unavailable from national surveillance systems. This novelty finding is further underscored due to drug detection rarely being reported from rural areas. These results are also timely and relevant, as data for Minnesota will not be available nationally for two more years. The most recent national data available for Minnesota only go through 2023 and do not provide nuance in toxicology to examine polydrug patterns (Kiang & Humphreys, 2025; Post et al., 2025). Despite these findings there are limitations, notably that the MMEO catchment area covers some, but not all of the state. Additionally, information about the concentration of substances detected in screening was unavailable and we were unable to detect all of the emerging fentanyl analogs with post-mortem toxicology results (Concheiro et al., 2018). Importantly, these results are not generalizable to the overall drug supply in the region as they represent only post-mortem toxicology from suspected overdose deaths and we cannot determine whether polydrug events were the result of co-use or contamination (Lockwood et al., 2021). Moreover, we cannot ascertain whether differences in polydrug use were the result of cultural preferences or drug distribution patterns (Hoebert et al., 2017; May & and Hough, 2004). Researchers were also unable to confirm the information on the death certificates and toxicology (race-ethnicity, military and marital status) which was collected as part of the death investigation process (Vakkalanka et al., 2025). Despite these limitations this study provides insight into public health outcomes in a highly rural region that otherwise would not be examined as part of inquiry into the ongoing overdose epidemic (Bensley et al., 2022; Carpenedo Mun et al., 2023; Coffey et al., 2020; Nolte et al., 2023; Staton et al., 2018). Importantly, the information examined in this study can allow these rural counties to evaluate the effectiveness of overdose prevention strategies in near real-time.

Finally, while our analysis was not designed to investigate causes of overdose reduction, we would be remiss in not describing the policies and practices occurring in the region. Minnesota has taken steps to expand access to evidence-based substance use disorder treatment including Medicaid reform, integrating services into primary care, increasing the number of prescribers, and supporting culturally responsive care through initiatives to enhance delivery in tribal communities (American Indian Team, 2025; Minnesota Department of Human Services, 2025; Minnesota Opioid Response, 2025). Importantly, Minnesota’s multifaceted public health approach emphasizes harm reduction as a core element, by supporting access to sterile syringes, widespread naloxone distribution, and the use of drug checking to reduce the health consequences of drug use (Minnesota Attorney General, 2025; Minnesota Department of Health, 2025a; Minnesota Department of Health, 2025b). These efforts to expand treatment capacity and harm reduction services have targeted rural communities, where geographic isolation and limited provider availability present persistent barriers to care. For example, by funding mobile harm reduction units, the state enables organizations to travel into the Iron Range to provide overdose prevention tools and linkage to evidence-based treatment (Palayew et al., 2023), all of which likely played a role in overdose reductions. Importantly if reductions in overdoses are attributable to these policies and practices, continued efforts are needed, particularly with an emphasis on culturally tailored overdose prevention strategies.

Conclusions

Collaborations among community partners to collect information on overdose events have been a critically important resource for understanding and responding to the evolving overdose crisis in the US. This study represents such an effort within a rural region, providing valuable insights into trends occurring among the populations that reside there, particularly those indigenous to the region. Future work will bring these novel toxicology data into action by monitoring emerging substances and informing strategies to reduce racial disparities in overdose outcomes.

Highlights.

• Overdose deaths declined by 40% in northern rural Minnesota from 2022 to 2024.

• American Indian overdose mortality remained 6 times higher than whites.

• Xylazine detected in 4.1% of overdose deaths, always with fentanyl present.