Overdose Deaths Involving Fentanyl and Fentanyl Analogs — New York City, 2000–2017

Cody Colon-Berezin, MSPH; Michelle L. Nolan, MPH; Jaclyn Blachman-Forshay, MPH; Denise Paone, EdD


Morbidity and Mortality Weekly Report. 2019;68(2):37-40. 

In This Article

Abstract and Introduction


Unintentional drug overdose deaths have climbed to record high levels, claiming approximately 70,000 lives in the United States in 2017 alone.[1] The emergence of illicitly manufactured fentanyl* (a synthetic, short-acting opioid with 50–100 times the potency of morphine) mixed into heroin, cocaine, and counterfeit pills, with or without the users' knowledge, has increased the risk for fatal overdose.[2,3] The New York City (NYC) Department of Health and Mental Hygiene (DOHMH) conducts routine overdose mortality surveillance by linking death certificates with toxicology findings from the NYC Office of the Chief Medical Examiner (OCME). A 55% increase in the rate of fatal drug overdose in NYC was observed from 2015 to 2017, resulting in the highest number of overdose deaths recorded since systematic reporting began in 2000. Toxicology data indicate that this unprecedented increase in overdose deaths is attributable to fentanyl. Early identification of increased fentanyl involvement enabled DOHMH to respond rapidly to the opioid overdose epidemic by increasing awareness of the risks associated with fentanyl and developing effective risk reduction messaging. These results strongly suggest that, wherever possible, jurisdictions should consider integrating toxicology findings into routine overdose surveillance and work with local medical examiners or coroners to include fentanyl in the literal text on death certificates.

Since 2013, illicitly manufactured fentanyl has been involved in a growing number of overdose deaths nationally[3,4] and in NYC and has been represented increasingly in seizures of synthetic opioids.[5,6] The increased presence of fentanyl in the illicit drug market has implications for overdose prevention efforts; however, national reporting on the presence of fentanyl in overdose deaths is limited by the lack of standardized toxicology testing for fentanyl and the inconsistent listing of fentanyl as a cause of death on death certificates, resulting in underreporting of fentanyl involvement in fatal overdoses. Nationally, reporting on drugs involved in overdose deaths relies on death certificate data; despite local efforts to improve drug reporting on death certificates, at least 15% of overdose deaths do not specify any drugs.[7] Thus, drug-specific data continue to be underreported, making it difficult to quantify the role of fentanyl in increasing overdose death rates.

For this analysis, DOHMH defined a death as an unintentional drug overdose if the medical examiner determined the manner of death to be accidental and the underlying or multiple-cause code was assigned an International Classification of Diseases, Tenth Revision code of X40–X44 (accidental overdose of drugs), F11–F16, or F18–F19 (excluding F-codes with 0.2 or 0.6 third digit). Toxicology findings were abstracted from OCME files and were used to classify overdose deaths according to the substances involved. Although OCME conducted fentanyl testing of all overdose cases during 2000–2012, universal testing for fentanyl was suspended during late 2013–July 2016, and the proportion of deaths tested during this time is unknown. However, despite inconsistent testing, in 2015 the proportion of all overdose deaths where fentanyl was detected exceeded that during the period of known universal fentanyl testing.

Unintentional drug overdose deaths were dichotomized according to whether or not any fentanyl was detected. The proportions of overdose deaths that involved fentanyl, overall and by other drug type involved, were calculated. Age-adjusted person-time rates were calculated by year using 2000–2016§ NYC population estimates adjusted to the U.S. Census 2000 projected population. Changes in rates were tested using z-tests and 95% confidence intervals; comparisons were based on the gamma confidence interval distribution.

Among 10,673 fatal overdoses in NYC during 2000–2014, a total of 7,822 (73%) involved an opioid. Fentanyl was determined to be involved in 246 of these deaths (i.e., 2% of all overdose deaths or 3% of deaths involving an opioid) (Figure). Beginning in 2015, the percentage of fentanyl-involved overdose deaths increased sharply; in 2016, 624 (44%) of 1,425 drug overdose deaths involved fentanyl, and in 2017, 842 (57%) of 1,487 overdose deaths involved fentanyl.


Number of overdose deaths and percentage of overdose deaths involving fentanyl* — New York City, 2000–2017
* Universal testing for fentanyl was stopped sometime during 2013 and restarted on July 1, 2016; fentanyl data during 2013–2016 were obtained from the Office of the Chief Medical Examiner but are known to be incomplete.

From 2014 to 2017, the rate of fentanyl-involved overdose deaths in NYC increased almost 3,000%, from 0.4 per 100,000 to 12.1. This trend is driving the overall increase in the rate of overdose deaths in NYC, which rose 81% during the same period, from 11.7 per 100,000 in 2014 to 21.2 in 2017, the highest rate since tracking of overdose deaths using this methodology began in 2000. In 2017, 531 (69%) heroin-involved deaths and 387 (53%) cocaine-involved overdose deaths also involved fentanyl (Table). Fentanyl also was involved in 146 (39%) deaths that involved cocaine but not heroin.

* In this report, "fentanyl" refers to both the pharmacologic compound and any of its analogs.
Disorders related to the use of opioids; cannabis; sedatives, hypnotics, or anxiolytics; cocaine; stimulants, hallucinogens; inhalants; or other psychoactive substances (excluding F-codes with 0.2 or 0.6 as the third digit, which specify a substance dependence or amnesic syndrome, respectively). DOHMH performs a secondary review on OCME case that have been assigned an X or F code because some F-coded cases involve acute drug intoxication. This review also facilitates reporting of toxicology involved.
§ DOHMH population estimates are modified from U.S. Census Bureau intercensal estimates.