The opioid overdose crisis continues to affect the lives of millions of people worldwide, with approximately 40.5 million people dependent on opioids and over 100,000 opioid-related overdose deaths in 2017 (1). The United States (US) is most highly impacted by the opioid overdose crisis (2–5). Over the past two decades, the opioid-related overdose death rate has increased by more than five-fold, from 2.9 per 100,000 in 1999 to 15.5 per 100,000 in 2019 (6), corresponding to a ten percent annual growth (7). Currently, New England is one of the most highly affected regions (8). In 2018, Massachusetts and Rhode Island had significantly greater opioid-related overdose death rates (29.3 per 100,000 and 25.9 per 100,000, respectively) than the national average of 15.5 per 100,000 (8).

Naloxone is a potent opioid antagonist that reverses the effects of an opioid-related overdose when administered by a bystander or health care professional (9). Federal, state, and local policies have been enacted to increase the distribution and availability of naloxone in communities across the US (10,11). However, there is significant heterogeneity in the penetration of naloxone into communities and evidence of racial disparities in access (12–14). This is partly due to healthcare facilities such as pharmacies and treatment centers being the primary distributor of naloxone in certain states (15,16). Compared to whites, non-white people who use drugs are less likely to have access to substance use treatment services due to economic and other structural barriers, and are more likely to reside in neighborhoods with limited access to pharmacies and pharmacies with an insufficient stock of naloxone (17–20). These factors—legacies of racist policies—result in decreased access to naloxone within non-white communities (19,21,22). Examples of these policies are the Anti-Drug Abuse Acts of 1986 and 1988 that disproportionately imprisoned African American/Black and Hispanic people, and mandated significantly longer sentences for crack/cocaine use as opposed to powder cocaine use (21). A 2016 study evaluated participant engagement in a community-based opioid overdose education and naloxone distribution (OEND) program and found that naloxone reversals were least likely to happen in census tracts with high proportions of African American residents, even though counts of opioid-related overdose deaths were highest in these same neighborhoods (12). Other studies have corroborated these findings, demonstrating that racially diverse neighborhoods have higher rates of overdose deaths yet do not have correspondingly high rates of naloxone utilization (13,14,23,24). The high rates of overdose deaths observed within non-white communities across the US demonstrate a need for increased coverage of and equitable access to naloxone in these settings. Thus, we aimed to determine if municipalities (cities, towns, or divisions of a place with its own local government) in Massachusetts and Rhode Island with relatively high proportions of non-white residents have inequitable access to naloxone distributed through community-based OEND programs.

To analyze the relationship between naloxone coverage equity and racial/ethnic composition of communities, we utilized health department naloxone data from Massachusetts and community-based naloxone data from Rhode Island from January 1st, 2016 to December 31st, 2018. Collectively, these data represent a census of all naloxone kits distributed through OEND organizations in both states. Based on an analysis conducted by one of the study’s co-authors (AC), annually, over 70% of naloxone dispensed in the study period was distributed from community-based OEND programs. The remaining amount were distributed through pharmacies. A recent study that evaluated the association between community characteristics and pharmacy-based naloxone distribution found that community racial/ethnic composition was not associated with naloxone dispensing outcomes in Rhode Island and Massachusetts (25). Due to the majority of naloxone being dispensed through OEND programs and the prior study’s finding of no difference in distribution by the racial composition of a community, we decided not to include pharmacy-based naloxone for this analysis. In Massachusetts, the major types of community-based OEND programs distributing naloxone are syringe service programs, which often collaborate with addiction treatment programs, and a statewide network of family support groups. As for Rhode Island, the major types are community education programs, treatment programs and syringe service programs.

Massachusetts’ OEND data is collected in collaboration with the Bureau of Infectious Disease and Laboratory Sciences and the Bureau of Substance Addiction Services. Twenty-seven medical and non-medical sites distribute naloxone funded through the Massachusetts Department of Public Health overdose prevention program (26,27). Rhode Island’s data collection process for naloxone distributed through community-based OEND organizations occurs through the Rhode Island Non-Pharmacy Naloxone Distribution program, centrally managed by the Rhode Island Department of Health (28). Before 2020, one of the largest OEND organizations in the state, Preventing Overdose and Naloxone Intervention, collected their own data on naloxone distributed from the organization. Thus, we used these two sources for naloxone data in Rhode Island for our study. Using these data sources, we calculated the annual counts of naloxone kits received by residents of each city/town.

We obtained opioid-related overdose deaths in each state (categorized by year and resident municipality of the person who died due to an opioid-related overdose) from the Office of State Medical Examiners and a publicly available report on opioid-related overdose deaths by municipality in Massachusetts (29). In each state, medical examiners determine drug type of overdose through toxicological analyses, death scene investigation, autopsy, and prior history of illegal drug use. An overdose death was determined as opioid-related if at least one of the following drugs were present: heroin, illicitly manufactured fentanyl, opioid-based prescription pain medications, and/or other unspecified opioids. In Rhode Island, only unintentional drug-related overdoses are reported and included in the analysis, whereas both intentional and unintentional overdoses were included in Massachusetts. Although Massachusetts included both types of overdoses, intentional cases accounted for less than five percent of all deaths; therefore, we do not believe this difference impacts the findings of the study in a meaningful way. Finally, both states have a single, centralized statewide medical examiner system (30,31). This results in all deaths being investigated in a consistent manner, which creates a highly reliable reporting system in both states.

The American Community Survey (ACS) was used to obtain percentage estimates of the population that self-identified as a race/ethnicity other than white (non-white) in each municipality (32). Racial/ethnic groups that were not white were collapsed into one category for the primary analysis due to many municipalities having less than one percent of residents identifying as either African American/Black, Hispanic, American Indian and Alaska Native, Asian, Native Hawaiian and Other Pacific Islander, or another race. Although there were concerns of small percentages of residents for various racial/ethnic sub-groups, we decided to conduct premeditated secondary analyses where we examined the percentage of African American/Black residents and the percentage of Hispanic/Latino residents as the independent variables of interest. We used the 5-year estimates released, for the year 2018, from ACS and assumed that the racial makeup of each jurisdiction was constant throughout the study period. Alongside race/ethnicity, we used the percentage of residents in poverty and urbanicity for each municipality from the ACS. The percentage of residents in poverty was determined by a threshold based on the household income and number of people in the family unit (33). If a household was under the threshold, they were considered to be in poverty. Due to the large range in population size (48–692,600) across municipalities, which impacts population density, we included urbanicity as a covariate in our models (32). For urbanicity, we used the US Census Bureau’s 2010 definition. A jurisdiction was considered urban if it had at least 1,000 people per square mile; everywhere else was considered non-urban (34,35). Those covariates were also assumed to be constant throughout the study period.

Three-hundred ninety municipalities were included in the analysis, with 351 in Massachusetts and 39 in Rhode Island. Across the two states, there were 26,616 kits distributed in 2016, 31,565 in 2017, and 35,965 in 2018. We excluded 4.8% (N=1,274) of naloxone kits in 2016 and 6 kits in 2017 due to missing data regarding the residing municipality of naloxone recipients. Over the study period, the naloxone coverage ratio among municipalities with at least one opioid-related overdose death varied between 0 and 135 with an annual mean municipality-level naloxone coverage ratio between 10.3 and 12.3 (Table 1). The scatter plots depicting the relationship between the naloxone coverage ratio and the percentage of non-white residents are shown in Figures 1 and 2. Each year of the study period, approximately 30% of the municipalities in Massachusetts surpassed the minimum. Of those exceeding the minimal threshold, 45–50 municipalities in the study period had an above-average (>13%) percentage of residents identifying as non-white. More municipalities crossed the minimum threshold over time in Rhode Island, with 8 in 2016 and increased to 27 in 2018. Of those with a ratio over 9, three municipalities had over 13% non-white residents in 2016, 8 in 2017, and 22 in 2018. Similar trends were observed when examining the relationships between naloxone coverage ratios and the percentage of African American/Black and Hispanic residents (see Supplemental Appendix).

The spatiotemporal distribution of naloxone coverage categories is summarized in Table 1 and visualized in Figures 3 and 4. In Massachusetts, most poorly resourced municipalities tended to be in the eastern region of the state (Figure 3). Wayland, MA was the only municipality that was consistently poorly resourced throughout the study period, with 18.8% non-white residents. We did not identify any consistently poorly resourced municipalities across all three years of the study in Rhode Island (Figure 4). Results were similar when the percentage of non-white residents was operationalized as quartiles.

The results of the bivariable analyses and multivariable models are summarized in Table 2. In the bivariable analysis, we found that urban municipalities, those with a higher poverty level, and higher percentages of African American/Black residents had greater naloxone coverage ratios, indicating increased naloxone distribution relative to overdose deaths. In the primary multivariable model (Model 1), there was no significant relationship between the percentage of non-white residents and the number of naloxone kits received by residents per opioid-related overdose death. In the secondary models (Model 2 and Model 3), the spline terms related to the percentage of African American/Black residents were found to be statistically significant, indicating that municipalities with higher percentages of African American/Black residents might have greater naloxone distribution relative to disease burden, after accounting for other community-level covariates.

In this multi-state study, we did not observe a relationship between naloxone coverage and municipality-level percentage of non-white residents. However, we observed a positive association between naloxone coverage and the percentage of African American/Black residents and residents in poverty. Moreover, of the 390 municipalities included in this study, only one was identified as being consistently poorly resourced relative to the state’s average naloxone coverage over the three-year study period. However, in both states, of the municipalities with a naloxone coverage ratio over 9 kits per overdose death, 43–50% of municipalities in Massachusetts and 37–40% in Rhode Island had a municipality-level percentage of non-white residents above the state average.

We found that municipalities with higher proportions of African American/Black residents had high numbers of naloxone kits distributed per opioid-related overdose death after accounting for other community-level characteristics. This is in contrast to recent studies that examined the geographic distribution of community-based naloxone reversals in San Francisco, California, where overdose reversals were least likely to occur in predominantly African American/Black neighborhoods (12) which could suggest poorer availability of naloxone in those communities. One likely explanation for our finding is that municipalities with high overdose rates in Massachusetts and Rhode Island also have above-average percentages of African American/Black residents (12,13,40–43). Because of the high overdose rates, more OEND organizations such as syringe service programs are likely located in these communities. This may result in higher naloxone distribution in these municipalities compared to other regions of the states. However, as this study is ecological, further research is needed to determine the extent to which naloxone is distributed equitably to racial/ethnic minorities residing in these municipalities.

Municipalities with higher percentages of residents in poverty also had higher rates of naloxone distribution through treatment and community programs supporting OEND. Like our findings for the percent of African American/Black residents, the association was in a different direction than other studies. A 2020 study using pharmacy-based naloxone distribution data found neighborhoods in New York City with lower income were least likely to have residents carrying naloxone (17). One of the likely causes of our differing findings is the type of distribution activity and the geographic scale. Early studies of pharmacy naloxone access found that pharmacies are least likely to have a sufficient supply of and stock naloxone in neighborhoods that are low-income, majority-minority, and have higher overdose death rates (17,18). In addition, in Massachusetts, a representative pharmacy purchase trial demonstrated that 81% of pharmacies provide naloxone access, but the study did not examine this access by race or ethnicity (44). In contrast, using data from state community-based naloxone distribution programs, we were able to examine how OEND organizations have increased naloxone distribution in low-income communities, particularly those with high opioid-related overdose deaths. In this manner, community-based OEND organizations may be effective at providing naloxone to higher-risk populations in communities with more non-white and low-income residents who continue to experience barriers while attempting to access naloxone from pharmacies and hospitals (12,17,19,22,45). As such, our findings suggest that OEND organizations potentially fill an important gap that is present for pharmacy naloxone distribution alone.

The study has several limitations. First, the measure we used to determine naloxone coverage equity may not accurately represent true access, particularly across communities of varying population sizes. Municipalities with low opioid-related overdose counts might be defined as poorly resourced due to the low demand for naloxone within the community. An example is Wayland, MA (with a population of less than 15,000), which was categorized as poorly resourced perhaps because the need for naloxone in the municipality was low, as evidenced by the fact that there were only two opioid-related overdose deaths that occurred during the entire study period. Second, while annualized naloxone coverage ratios permitted us to determine changes in distribution patterns over time, we were unable to assess the precise temporality of naloxone distribution activities compared to opioid-related overdose deaths. Another limitation is the geographic unit chosen for the analysis. There might be a difference in outcomes if we used a smaller geographic unit such as zip codes or census tracts that may be more highly correlated with specific racial and ethnic populations. Also, our study does not consider the impact of pharmacy distribution of naloxone. Both pharmacies and community-based organizations are increasing their naloxone distribution among people at risk of experiencing an overdose. This leads to an interconnected impact on the overdose rates that were used in the study. Lastly, our data did not permit the determination of whether a naloxone kit was used for a reversal. Therefore, we have no knowledge of how many of the naloxone kits that were distributed were used and where they were used.

In conclusion, after taking into account opioid-related overdose mortality burden and other factors, we did not identify municipal-level racial/ethnic inequities in naloxone distribution in Rhode Island and Massachusetts. While this finding is encouraging, we were unable to assess the racial/ethnic identity of people within those municipalities who are receiving naloxone kits and whether it reflects the racial makeup of persons who are most likely to experience an overdose. Future studies should be conducted to evaluate individual- and structural-level barriers to naloxone access, particularly those experienced by racial/ethnic minority residents in municipalities with high opioid overdose burden.