Abstract
Objectives. To examine the impact of North Carolina’s 2017 Strengthening Opioid Misuse Prevention (STOP) Act on opioid overdose deaths.
Methods. We used quarterly data from the North Carolina Opioid Dashboard to conduct an interrupted time series analysis ranging from 2010 to 2018. Results were stratified by heroin–fentanyl deaths and other opioid deaths.
Results. After the STOP Act, there was an initial rate increase of 0.60 opioid deaths per 100 000 population (95% confidence interval [CI] = 0.04, 1.15) and a decrease of 0.42 (95% CI = −0.56, −0.29) every quarter thereafter. Results differed by stratification.
Conclusions. Our results suggest that North Carolina’s STOP Act was associated with a reduction in opioid deaths in the year following enactment. The changes in opioid overdose death trends coinciding with the STOP Act were similar to outcomes seen with previous opioid policies.
Public Health Implications. Future policies designed to reduce the availability of opioids may benefit from encouraging and increasing the availability of evidence-based treatment of opioid use disorder.
In North Carolina, the ongoing impact of opioids resulted in more than 13 000 deaths between 1999 and 2017.1,2 In response, the Strengthening Opioid Misuse Prevention (STOP) Act passed the North Carolina General Assembly unanimously and was signed into law in June 2017, implementing rolling changes to regulations through 2020.3 The first 2 waves of policy changes were enacted in the third quarter of 2017 (July and September, respectively). These initial 2 waves implemented regulations that required physician assistants and nurse practitioners in pain clinics to consult with the supervising physician for any opioid prescription of more than 30 days, as well as every 90 days that the prescription remained active (wave 1).
In addition, pharmacies were required to report opioid prescriptions to the Controlled Substance Reporting System (North Carolina’s version of a prescription drug monitoring program) no more than 1 day after they were dispensed, with potential monetary penalties assessed by the North Carolina Department of Health and Human Services (DHHS) if reports were found to be missing or incomplete (wave 2). Finally, the STOP Act sought to increase access to naloxone at the community level in a number of ways. For example, it lifted a state ban on the use of local funds for syringe access programs (although it did not provide such funding; wave 1). Also, it expanded North Carolina’s Good Samaritan Law4 by allowing providers to facilitate government and nonprofit organizations’ ability to obtain and distribute naloxone via a standing order and by providing those organizations with limited immunity otherwise previously offered only to individuals (wave 1).
Research has demonstrated inconsistencies regarding the efficacy of opioid policies in reducing overdose deaths, including some unintended outcomes. Policies intending to restrict the accessibility of prescription opioids have been effective in reducing prescription opioid overdose deaths, but they have also been associated with increased deaths from heroin and fentanyl. Conversely, policies focused on treating substance use disorders are typically effective in reducing all opioid overdose deaths.5–8 Given that the effects of the STOP Act were unknown, we explored the impact of the first 2 waves of the act on opioid-related deaths.
METHODS
We used publicly available data from the North Carolina Opioid Dashboard to assess the impact of the STOP Act on opioid-related deaths.9 This dashboard, constructed and maintained by the North Carolina Department of Health and Human Services as part of the state’s opioid action plan,10 tracks several opioid-related metrics on a quarterly basis. Data relating to opioid deaths were derived from North Carolina DHHS death certificate information; this information, tracked by the agency’s Vital Statistics Office, represents the official count for the state.9 The dashboard was updated with new quarterly data beginning in 2017, with prior years of DHHS data being included for most factors going back to 2010, resulting in a total possible sample of 36 quarters over 9 years (2010–2018). Data from 2010 through 2018 were obtained in December 2019.
Opioid-Related Deaths
Opioid-related deaths, measured as a count of all unintentional opioid-related deaths among North Carolina residents as defined by International Classification of Diseases, 10th Revision (ICD-10; World Health Organization, Geneva, Switzerland), codes X40 to X44 (with T codes of 40.0 to 40.4 or 40.6), were available quarterly from 2010 to 2018 (n = 36).9 Opioid-related deaths were stratified into heroin- and fentanyl-related deaths and nonheroin- and nonfentanyl-related deaths for our subanalyses. We created these strata by multiplying the total opioid-related death count by the percentage of all opioid-related deaths that specifically involved heroin, fentanyl, or fentanyl analogs (acetyl fentanyl, butrylfentanyl, furanylfentanyl, fluorofentanyl, acrylfentanyl, fluoroisobutrlfentanyl, betahydroxythiofentanyl, carfentanil).9
Data on percentages of deaths involving heroin, fentanyl, or fentanyl analogs were available yearly from 2010 to 2016 and quarterly from 2016 to 2018. We filled in quarterly percentages for 2010 to 2016 by applying the yearly percentage to each quarter within the year. For example, the 2015 annual percentage of 46.8% resulted in each quarter in 2015 having a value of 46.8%. We converted all 3 measures of death to unadjusted rates per 100 000 using the annual population estimates available through the North Carolina Opioid Dashboard.
STOP Act
A binary indicator variable was created that coded each quarter as being either before or after implementation of the STOP Act, with quarter 3 of 2017 (July–September) marked as the first time period for implementation. We selected quarter 3 because the resulting changes to state-level prescribing guidelines that took effect in July 2017 have been cited as motivating many physicians in North Carolina to immediately and abruptly stop prescribing opioids entirely, including for chronic pain.11
With previous evidence suggesting that such a decrease or discontinuation of opioid prescribing can contribute to an increase in heroin use, including a warning issued by the US Food and Drug Administration, we sought to capture any initial increases in heroin deaths that may have been associated with the first wave of the STOP Act.12–15 However, several portions of the act may require a “phase-in” period (e.g., effects of changes to the Good Samaritan Law). To examine this possibility, we conducted a sensitivity analysis in which quarter 3 of 2017 was suppressed and quarter 4 served as the first postimplementation period.
Data Analysis
Interrupted time series analysis, a quasi-experimental method, was used to estimate the impact of the STOP Act on opioid-related deaths.16 We used the full 36 quarters of data to generate the most accurate estimate of the underlying temporal trends before implementation of the act. The model contained 2 time components to account for these underlying temporal trends: a continuous time variable from 1 to 36 marking each of the quarters from 2010 to 2018 to model linear temporal trends and a time squared component to accommodate nonlinear temporal trends.
The binary variable representing STOP Act implementation (coded as 0 before implementation and 1 after implementation) was included in the model, as was a variable for time after STOP Act implementation (coded as 0 before the STOP Act and as a continuous variable [1–5] for the quarters after implementation). This allowed us to estimate both the immediate impact of the STOP Act (using the coefficient of the binary STOP Act variable) and the change in temporal trends after implementation (using the time after implementation variable) while accounting for preexisting temporal trends.
We conducted our analysis using PROC AUTOREG in SAS version 9.4 (SAS Institute Inc, Cary, NC) to accommodate the autocorrelation that can exist in time series data. As our data were quarterly, we tested for autocorrelation using Durbin–Watson statistics for orders 1 through 5. To identify the most parsimonious model while still accounting for autocorrelation, we used a stepwise autoregression for the overall analysis and each of the stratified analyses. This technique incorporates backward parameter selection to identify the order of the autoregression.17–20 A maximum order of 5 was allowed, and nonsignificant autoregressive parameters were deleted until only significant (P < .05) parameters remained.
RESULTS
In the analysis of the overall opioid death rate, no autoregressive parameters were significant at a P level of less than .05, and therefore no adjustments were made to the model to address autocorrelation. The model results indicate that, after adjustment for time trends, there was an initial increase of 0.60 deaths per 100 000 population in the total opioid-related mortality rate (95% confidence interval [CI] = 0.04, 1.15) during the quarter the STOP Act took effect and a decrease of 0.42 in the quarterly estimated temporal trend after implementation (95% CI = −0.56, −0.29; Figure 1). In our sensitivity analysis assessing the overall opioid death rate (in which data from quarter 3 of 2017 were suppressed and quarter 4 served as the first postimplementation period), the STOP Act was not associated with an initial increase (−0.06; 95% CI = −0.50, 0.37), although the quarterly decrease remained (−0.35; 95% CI = −0.48, −0.22; Appendix A, available as a supplement to the online version of this article at http://www.ajph.org).
FIGURE 1—
Impact of the Strengthening Opioid Misuse Prevention (STOP) Act on Opioid-Related Death Rates Estimated Through Interrupted Time Series Analysis: North Carolina, 2010–2018
Note. The vertical dotted line represents implementation of waves 1 and 2 of the STOP Act (in quarter 3 of 2017). The counterfactual (long dashed line) was estimated by setting the effects of the STOP Act to 0 and estimating the death rate adjusting for time and time squared. The death rate increased in the same quarter as the STOP Act (0.60 deaths per 100 000 population; 95% confidence interval [CI] = 0.04, 1.15) and declined every quarter after (−0.42 deaths per 100 000 population; 95% CI = −0.56, −0.29).
In the stratified analysis, the nonheroin–nonfentanyl model differed in that the STOP Act was associated with an initial decrease in the nonheroin–nonfentanyl death rate (−0.28 per 100 000 population; 95% CI = −0.48, −0.08); however, there was not a significant reduction in the quarterly temporal trend (−0.04, 95% CI = −0.09, 0.02; Figure 2). The model for heroin–fentanyl deaths was similar to that in the unstratified analysis, showing that the STOP Act was associated with an initial increase of 1.07 per 100 000 population in the heroin–fentanyl death rate (95% CI = 0.65, 1.48). After the initial spike, there was a reduction of 0.42 in the quarterly temporal trend (95% CI = −0.54, −0.30; Figure 3). (Table 1 presents estimates and confidence intervals.) The sensitivity analysis for stratified overdose deaths yielded similar patterns in terms of both significance and directionality (sensitivity analysis graphs are provided in Appendices B and C, and sensitivity analysis estimates, confidence intervals, and associated P values are provided in Appendix D, available as supplements to the online version of this article at http://www.ajph.org).
FIGURE 2—
Impact of the Strengthening Opioid Misuse Prevention (STOP) Act on Nonheroin- and Nonfentanyl-Related Death Rates Estimated Through Interrupted Time Series Analysis: North Carolina, 2010–2018
Note. The vertical dotted line represents implementation of waves 1 and 2 of the STOP Act (in quarter 3 of 2017). The counterfactual (long dashed line) was estimated by setting the effects of the STOP Act to 0 and estimating the death rate adjusting for time and time squared. The death rate decreased in the same quarter as the STOP Act (−0.28 deaths per 100 000 population; 95% confidence interval [CI] = −0.48, −0.08). The rate did not involve a change in trend associated with the STOP Act (−0.04 deaths per 100 000 population; 95% CI = −0.09, 0.02).
FIGURE 3—
Impact of the Strengthening Opioid Misuse Prevention (STOP) Act on Heroin- and Fentanyl-Related Death Rates Estimated Through Interrupted Time Series Analysis: North Carolina, 2010–2018
Note. The vertical dotted line represents implementation of waves 1 and 2 of the STOP Act (in quarter 3 of 2017). The counterfactual (long dashed line) was estimated by setting the effects of the STOP Act to 0 and estimating the death rate adjusting for time and time squared. The death rate increased in the same quarter as the STOP Act (1.07 deaths per 100 000 population; 95% confidence interval [CI] = 0.65, 1.48) and declined every quarter thereafter (−0.42 deaths per 100 000 population; 95% CI = −0.54, −0.30).
TABLE 1—
Interrupted Time Series Estimates of the Effects of the Strengthening Opioid Misuse Prevention (STOP) Act: North Carolina, 2010–2018
| Total Death Rate, Estimate (95% CI) | Nonheroin–Nonfentanyl Death Rate, Estimatea (95% CI) | Heroin–Fentanyl Death Rate, Estimateb (95% CI) | |
| Intercept | 2.070 (1.758, 2.371) | 1.425 (1.266, 1.585) | 0.541 (0.199, 0.883) |
| STOP Act | 0.596 (0.044, 1.148) | −0.281 (−0.479, −0.083) | 1.068 (0.654, 1.482) |
| Time after STOP Act | −0.421 (−0.557, −0.286) | −0.035 (−0.093, 0.024) | −0.415 (−0.536, −0.295) |
| Time | −0.094 (−0.140, −0.049) | −0.004 (−0.028, 0.019) | −0.078 (−0.126, −0.027) |
| Time × Time | 0.006 (0.004, 0.007) | 0.000 (−0.001, 0.001) | 0.005 (0.004, 0.007) |
Note. CI = confidence interval.
Yule–Walker estimates are adjusted for autoregressive lags of 1, 2, and 4.
Yule–Walker estimates are adjusted for an autoregressive lag of 2.
DISCUSSION
We modeled the effects of the STOP Act on opioid-related deaths in North Carolina. Our results provide some evidence that the STOP Act was associated with an initial increase in the death rate followed by a steady decline. This suggests that the impact of the policy is a successful reduction in the rate of opioid-related deaths, although future studies should examine the policy’s practical implementation and long-term effects.
The results of a sensitivity analysis allowing for a quarter-long implementation period did not reveal an initial increase in the death rate; however, this more conservative analysis did yield a quarterly rate decrease consistent with the full analysis. After stratification into heroin–fentanyl deaths and nonheroin–nonfentanyl deaths, it seems that the STOP Act may have affected these 2 types of deaths differently. In the case of heroin–fentanyl deaths, the STOP Act was associated with an initial increase in the death rate followed by a decrease over time. Conversely, the nonheroin–nonfentanyl death rate declined in the period immediately following the STOP Act, although it did not exhibit a decrease in the quarterly trend. This lack of a trend decrease may have been a result of the rate of nonheroin–nonfentanyl deaths already being noticeably lower than the overall death rate and the heroin–fentanyl rate. The more conservative estimates of the stratified effects calculated in the sensitivity analysis were consistent with both the initial and quarterly trend estimates of the full analysis.
These early changes in overdose deaths associated with the STOP Act appear to be consistent with the results of other opioid policies in several ways. Supply side policies, such as the increased prescribing regulations in the STOP Act, have been conjectured to lead to fewer deaths from prescription opioids at the expense of increases in heroin–fentanyl deaths.7,8,21–25
A 2018 survey conducted by the North Carolina Medical Society revealed that more than 600 providers in North Carolina had stopped prescribing opioids, with some citing the STOP Act as the reason.11 This reduction in providers prescribing opioids, whether specifically a result of the STOP Act or to increased general awareness associated with the passing of the act, could have led to an increase in new or relapsed users of heroin–fentanyl as well as a possible increase in the number of people with opioid use disorder not receiving treatment, which carries an increased risk of overdose.26 Conversely, demand-side policies, such as the standing order for naloxone included in the STOP Act, have been associated with decreases in both prescription opioid overdose deaths and heroin–fentanyl overdose deaths.6,8,26,27 The STOP Act, accordingly, was associated with an initial decrease in prescription deaths and decreases in all opioid overdose deaths over time.
It is possible that the initial increase in opioid deaths shown in our results was not related to the STOP Act but, rather, attributable to other extraneous factors. The elevated heroin- and fentanyl-related death rate for quarter 3 of 2017 may have been a natural continuation of the exponential growth seen in previous quarters. Furthermore, it appears that there were regional trends during the observation period. Vivolo-Kantor et al. found that emergency department visits for suspected opioid overdoses spiked between quarters 2 and 3 of 2017 in the southeastern United States, following a pattern similar to that seen in our analysis and possibly accounting for some of the increase.28 Future studies should examine patient and provider experiences relating to the implementation of the STOP Act, the opioid prescribing guidelines of the Centers for Disease Control and Prevention (CDC), and related policies in an attempt to better understand the effects on opioid-related overdoses and deaths. Future studies should also examine the long-term effects of the STOP Act.
Limitations
Our study had several limitations. In particular, although the quasi-experimental nature of interrupted time series analysis is robust, our use of observational data precludes us from truly assessing causality. As a result of the urgency of opioid overdose trends and ongoing efforts to address them, there is the potential for history effects in our study resulting from national and local initiatives. For example, during the time frame in which our data were collected and prior to the STOP Act, the CDC issued national guidelines on opioid prescribing.29
There is also the potential for confounding in this study owing to heightening general awareness regarding opioid overdose risks. It is possible that an increase in public awareness contributed to fewer deaths as more people became aware of effective treatment options for opioid use disorder, as well as overdose reversal medication (naloxone), and more people accessed relevant services earlier. Although we cannot entirely rule out the possibility of such history effects, there is some evidence that history effects may be minimal. Tennessee and South Carolina, neighboring states similar to North Carolina in not expanding Medicaid, saw an increase in opioid deaths in 2018.30,31
Our outcome measure was based on state-compiled death certificate data, which are not collected for research purposes. Death certificates are known to involve inaccuracies in reporting, especially as it relates to overdose deaths, which may have affected our results.32–34 In addition, the percentage of opioid overdose deaths attributable to heroin and fentanyl was reported annually from 2010 to 2016, which may have obscured some quarterly variation within those years. However, it is worth noting that the quarterly percentages available appeared tightly clustered around the annual percentage. The use of state-level data also limited our ability to make inferences about specific regions within the state.
Finally, we did not consider the effects of local factors, such as socioeconomic status, urbanicity, regional variations in policy compliance and implementation, and local policies or service availability (such as syringe access programs that distribute naloxone), that may have further affected overdose death rates. An example is the potential for shifting trends in substances of misuse. There is evidence to suggest that other substances are rising in use. For example, 1 study showed that methamphetamine and opioid co-use doubled between 2011 and 2017.35
Public Health Implications
Our results suggest that there were decreases in opioid deaths associated with implementation of the STOP Act. This is consistent with previous literature in that it offers support for the efficacy of multifaceted state-level opioid policies. However, also consistent with previous literature was the initial increase in heroin–fentanyl deaths. We propose that future opioid deprescribing policies incorporate additional components encouraging and increasing access to evidence-based treatment of opioid use disorder.
CONFLICTS OF INTEREST
The authors report no conflicts of interest.
HUMAN PARTICIPANT PROTECTION
No protocol approval was needed for this study because no human participants were involved.
Footnotes
See also Cooper, p. 1450.
REFERENCES
- 1.Ostrach B, Hayes V. Overdose deaths: not an epidemic or a crisis, and not by accident. Available at: https://www.thefix.com/causes-of-increasing-overdose-deaths. Accessed June 5, 2020.
- 2.North Carolina Department of Health and Human Services. Injury and Violence Prevention Branch. The opioid-related overdose fact sheet. Available at: https://www.injuryfreenc.ncdhhs.gov/DataSurveillance/Poisoning.htm. Accessed June 5, 2020.
- 3.General Assembly of North Carolina. STOP Act. Available at: https://www.ncmedboard.org/images/uploads/article_images/H243v7.pdf. Accessed June 5, 2020.
- 4.North Carolina Harm Reduction Coalition. Overdose prevention law in North Carolina. Available at: http://www.nchrc.org/programs-and-services/911-good-samaritan-laws-naloxone-access-and-syringe-law-in-nc. Accessed June 5, 2020.
- 5.Roberts AW, Farley JF, Holmes GM et al. Controlled substance lock-in programs: examining an unintended consequence of a prescription drug abuse policy. Health Aff (Millwood) 2016;35(10):1884–1892. doi: 10.1377/hlthaff.2016.0355. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Wakeland W, Nielsen A, Schmidt TD. Gaining policy insight with a system dynamics model of pain medicine prescribing, diversion and abuse: pain medicine policy model. Syst Res. 2016;33(3):400–412. [Google Scholar]
- 7.Kertesz SG, Gordon AJ. A crisis of opioids and the limits of prescription control: United States. Addiction. 2019;114(1):169–180. doi: 10.1111/add.14394. [DOI] [PubMed] [Google Scholar]
- 8.Pitt AL, Humphreys K, Brandeau ML. Modeling health benefits and harms of public policy responses to the US opioid epidemic. Am J Public Health. 2018;108(10):1394–1400. doi: 10.2105/AJPH.2018.304590. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.North Carolina Department of Health and Human Services. Opioid dashboard. Available at: https://injuryfreenc.shinyapps.io/OpioidActionPlan. Accessed June 5, 2020.
- 10.North Carolina Department of Health and Human Services. Opioid action plan. Available at: https://www.ncdhhs.gov/about/department-initiatives/opioid-epidemic/north-carolinas-opioid-action-plan. Accessed June 5, 2020.
- 11.Knopf T. Hundreds of N.C. doctors say they’ve stopped prescribing opioids. Available at: https://www.northcarolinahealthnews.org/2018/10/15/nc-doctors-stop-prescribe-opioids. Accessed June 5, 2020.
- 12.US Food and Drug Administration. FDA identifies harm reported from sudden discontinuation of opioid pain medicines and requires label changes to guide prescribers on gradual, individualized tapering. Available at: https://www.fda.gov/drugs/drug-safety-and-availability/fda-identifies-harm-reported-sudden-discontinuation-opioid-pain-medicines-and-requires-label-changes. Accessed June 5, 2020.
- 13.Cicero TJ, Ellis MS, Surratt HL. Effect of abuse-deterrent formulation of OxyContin. N Engl J Med. 2012;367(2):187–189. doi: 10.1056/NEJMc1204141. [DOI] [PubMed] [Google Scholar]
- 14.Compton WM, Jones CM, Baldwin GT. Relationship between nonmedical prescription-opioid use and heroin use. N Engl J Med. 2016;374(2):154–163. doi: 10.1056/NEJMra1508490. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Mars SG, Bourgois P, Karandinos G, Montero F, Ciccarone D. “Every ‘never’ I ever said came true”: transitions from opioid pills to heroin injecting. Int J Drug Policy. 2014;25(2):257–266. doi: 10.1016/j.drugpo.2013.10.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Penfold RB, Zhang F. Use of interrupted time series analysis in evaluating health care quality improvements. Acad Pediatr. 2013;13(6):S38–S44. doi: 10.1016/j.acap.2013.08.002. [DOI] [PubMed] [Google Scholar]
- 17.Miller B, Kassenborg H, Dunsmuir W et al. Syndromic surveillance for influenzalike illness in ambulatory care setting. Emerg Infect Dis. 2004;10(10):1806–1811. doi: 10.3201/eid1010.030789. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Ho H-M, Rao CY, Hsu H-H, Chiu Y-H, Liu C-M, Chao HJ. Characteristics and determinants of ambient fungal spores in Hualien, Taiwan. Atmos Environ. 2005;39(32):5839–5850. [Google Scholar]
- 19.Nagata T, Setoguchi S, Hemenway D, Perry MJ. Effectiveness of a law to reduce alcohol-impaired driving in Japan. Inj Prev. 2008;14(1):19–23. doi: 10.1136/ip.2007.015719. [DOI] [PubMed] [Google Scholar]
- 20.Moineddin R, Upshur RE, Crighton E, Mamdani M. Autoregression as a means of assessing the strength of seasonality in a time series. Popul Health Metr. 2003;1(1):10. doi: 10.1186/1478-7954-1-10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Beletsky L, Davis CS. Today’s fentanyl crisis: prohibition’s iron law, revisited. Int J Drug Policy. 2017;46:156–159. doi: 10.1016/j.drugpo.2017.05.050. [DOI] [PubMed] [Google Scholar]
- 22.Cicero TJ, Ellis MS. Understanding the demand side of the prescription opioid epidemic: does the initial source of opioids matter? Drug Alcohol Depend. 2017;173:S4–S10. doi: 10.1016/j.drugalcdep.2016.03.014. [DOI] [PubMed] [Google Scholar]
- 23.Cicero TJ, Ellis MS, Kasper ZA. Increased use of heroin as an initiating opioid of abuse. Addict Behav. 2017;74:63–66. doi: 10.1016/j.addbeh.2017.05.030. [DOI] [PubMed] [Google Scholar]
- 24.Cicero TJ, Kasper ZA, Ellis MS. Increased use of heroin as an initiating opioid of abuse: further considerations and policy implications. Addict Behav. 2018;87:267–271. doi: 10.1016/j.addbeh.2018.05.030. [DOI] [PubMed] [Google Scholar]
- 25.Hadland SE, Beletsky L. Tighter prescribing regulations drive illicit opioid sales. BMJ. 2018;361:k2480. doi: 10.1136/bmj.k2480. [DOI] [PubMed] [Google Scholar]
- 26.Clark RE, Samnaliev M, Baxter JD, Leung GY. The evidence doesn’t justify steps by state Medicaid programs to restrict opioid addiction treatment with buprenorphine. Health Aff (Millwood) 2011;30(8):1425–1433. doi: 10.1377/hlthaff.2010.0532. [DOI] [PubMed] [Google Scholar]
- 27.Fatseas M, Auriacombe M. Why buprenorphine is so successful in treating opiate addiction in France. Curr Psychiatry Rep. 2007;9(5):358–364. doi: 10.1007/s11920-007-0046-2. [DOI] [PubMed] [Google Scholar]
- 28.Vivolo-Kantor AM, Seth P, Gladden RM et al. Vital signs: trends in emergency department visits for suspected opioid overdoses—United States, July 2016–September 2017. MMWR Morb Mortal Wkly Rep. 2018;67(9):279–285. doi: 10.15585/mmwr.mm6709e1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Centers for Disease Control and Prevention. Guideline for prescribing opioids for chronic pain. Available at: https://www.cdc.gov/drugoverdose/pdf/Guidelines_Factsheet-a.pdf. Accessed June 5, 2020. [DOI] [PubMed]
- 30.Burch K. Tennessee overdose rates hit five-year high. Available at: https://www.thefix.com/tennessee-overdose-rates-hit-five-year-high. Accessed June 5, 2020.
- 31.South Carolina Department of Health and Environmental Control. Drug overdose deaths: South Carolina. Available at: https://www.scdhec.gov/health/sc-public-health-statistics-maps/biostatistics-publications. Accessed June 5, 2020.
- 32.Buchanich JM, Balmert LC, Williams KE, Burke DS. The effect of incomplete death certificates on estimates of unintentional opioid-related overdose deaths in the United States, 1999–2015. Public Health Rep. 2018;133(4):423–431. doi: 10.1177/0033354918774330. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Ruhm CJ. Corrected US opioid-involved drug poisoning deaths and mortality rates. Addiction. 1999–2015;2018;113(7):1339–1344. doi: 10.1111/add.14144. [DOI] [PubMed] [Google Scholar]
- 34.Mertz KJ, Janssen JK, Williams KE. Underrepresentation of heroin involvement in unintentional drug overdose deaths in Allegheny County, PA. J Forensic Sci. 2014;59(6):1583–1585. doi: 10.1111/1556-4029.12541. [DOI] [PubMed] [Google Scholar]
- 35.Ellis MS, Kasper ZA, Cicero TJ. Twin epidemics: the surging rise of methamphetamine use in chronic opioid users. Drug Alcohol Depend. 2018;193:14–20. doi: 10.1016/j.drugalcdep.2018.08.029. [DOI] [PubMed] [Google Scholar]