BACKGROUND: Unintentional opioid overdose death is a national epidemic. Since the passage of the North Carolina SB20 in 2013, two-dose naloxone rescue kits have become more widely available via enhanced naloxone distribution throughout NC. We conducted an analysis comparing high-naloxone receiving counties to the rest of the state, with the intent to assess the enhanced distribution’s impact on mortality.
METHODS: We calculated the number of two-dose naloxone rescue kits distributed in each county by the NC Harm Reduction Coalition during 2013−2014. Unintentional opioid overdose deaths by county from 1999-2014 were calculated using vital records and census data. We classified five counties as high distribution counties and also calculated counties in the 90th percentile for distribution. Counties with high death rates and low naloxone distribution were classified as control counties. Annual unintentional opioid overdose death rates for high distribution counties and 90thpercentile counties were compared to NC rates over time using methods (chi-square test and ANOVA). We calculated and compared the annual percent change in death rates between high distribution and control counties using a chi-square test. We compared 1999−2012 rates in high distribution counties to rates in the same counties for 2013−2014 using a paired T test.
RESULTS: Sixty-three percent of naloxone rescue kits were distributed to counties in the 90th percentile of distribution. No statistically significant differences in death rate were found between high distribution counties and NC by chi square (p-value 0.76) or ANOVA (0.2561), or between counties in the 90thpercentile of distribution and the rest of NC by chi square (p-value 0.9563) or ANOVA (p-value 0.86). Likewise, no statistically significant difference was found when comparing the percent change in death rates between high distribution counties and control counties by chi square (p-value XX) Using a paired T-test to compare rates in high distribution counties before and after enhanced distribution began, we observed a statistically significant difference among the 90th percentile group (p=0.01), with a higher rate observed in the latter time period. This difference was not detected when looking at the rest of NC (p-value 0.07).
CONCLUSIONS: This suggests the counties targeted to receive the greatest volume of naloxone rescue kits are the counties seeing the greatest rise in deaths. This work highlights the complexity of quantifying the impact of naloxone distribution. We are engaging with partners to discuss the additional work needed to identify appropriate methods for analyzing the available data to assess trends overtime.