Morbidity and Mortality Weekly Report

Weekly / Vol. 68 / No. 34 August 30, 2019

International Overdose Awareness Changes in Opioid-Involved Overdose

Day — August 31, 2019 Deaths by Opioid Type and Presence
of Benzodiazepines, Cocaine, and
August 31, 2019, is International Overdose Awareness
Day, a global event that aims to raise awareness that over- Methamphetamine — 25 States, July–
dose death is preventable and to reduce the stigma associ- December 2017 to January–June 2018
ated with drug-related death. Goals also include providing
information about risk for overdose and community services R. Matt Gladden, PhD1; Julie O’Donnell, PhD1;
and preventing drug-related harm through evidence-based Christine L. Mattson, PhD1; Puja Seth, PhD1
policy and practice (https://www.overdoseday.com). From 2013 to 2017, the number of opioid-involved overdose
The opioid overdose epidemic, which killed 47,600 U.S. deaths (opioid deaths) in the United States increased 90%,
persons in 2017,* substantially expanded in 2013 driven from 25,052 to 47,600.* This increase was primarily driven by
by rapid increases in overdose deaths involving synthetic substantial increases in deaths involving illicitly manufactured
opioids (excluding methadone), particularly illicitly fentanyl (IMF) or fentanyl analogs† mixed with heroin, sold
manufactured fentanyl.† Cocaine and methamphetamine as heroin, or pressed into counterfeit prescription pills (1–3).
overdose deaths co-involving synthetic opioids also rapidly Methamphetamine-involved and cocaine-involved deaths that
increased during this period (1).
A report in this issue of MMWR documented decreases * https://www.cdc.gov/nchs/data/databriefs/db329_tables-508.pdf#4.
† Fentanyl is a synthetic opioid 50–100 times more potent than morphine and
in opioid-involved overdose deaths in 25 states from July–
is approved for treatment of severe (typically advanced cancer) pain. Illicitly
December 2017 to January–June 2018, especially those manufactured fentanyl is manufactured illegally and sold through illegal drug
involving fentanyl analogs and prescription opioids. Overdose markets for its heroin-like effect. Fentanyl analogs, also known as fentanyl-
related substances, are synthetic opioids that are similar in chemical structure
deaths involving illicitly manufactured fentanyl (includ- to fentanyl but modified to generate distinct substances. Fentanyl analogs vary
ing those co-occurring with illicit opioids and stimulants) in potency, with some more potent than fentanyl and others with potency
increased (2). Improved identification of persons at high risk similar to or less than fentanyl. https://www.cdc.gov/drugoverdose/opioids/
fentanyl.html; https://www.deadiversion.usdoj.gov/drug_chem_info/frs.pdf.
for overdoses involving illicitly manufactured fentanyl and
linkage to risk-reduction services and evidence-based treat-
ment are critical to reducing opioid deaths. Further informa-
INSIDE
tion on CDC’s state efforts and overdose data is available at
https://www.cdc.gov/drugoverdose/index.html. 745 Racial Disparities in Breastfeeding Initiation and
Duration Among U.S. Infants Born in 2015
* https://www.cdc.gov/drugoverdose/data/statedeaths.html. 749 Notes from the Field: Mumps in Detention Facilities
† https://www.cdc.gov/drugoverdose/epidemic/index.html. that House Detained Migrants — United States,
September 2018–August 2019
References 751 Notes from the Field: Multistate Outbreak of
1. Kariisa M, Scholl L, Wilson N, Seth P, Hoots B. Drug overdose deaths Salmonella Agbeni Associated with Consumption of
involving cocaine and psychostimulants with abuse potential—United Raw Cake Mix — Five States, 2018
States, 2003–2017. MMWR Morb Mortal Wkly Rep 2019;68:388–95.
https://doi.org/10.15585/mmwr.mm6817a3 753 QuickStats
2. Gladden RM, O’Donnell J, Mattson C, Seth P. Changes in opioid-
involved overdose deaths by opioid type and presence of
benzodiazepines, cocaine, and methamphetamine—25 states, July– Continuing Education examination available at
December 2017 to January–June 2018. MMWR Morb Mortal Wkly https://www.cdc.gov/mmwr/cme/conted_info.html#weekly.
Rep 2019;68:737–44.

U.S. Department of Health and Human Services

Centers for Disease Control and Prevention
 Morbidity and Mortality Weekly Report

co-involved opioids also substantially increased from 2016 to 2017 presence (detection of the drug in decedent) of co-occurring
(4). Provisional 2018§ estimates of the number of opioid deaths nonopioid drugs (cocaine, methamphetamine, and benzodi-
suggest a small decrease from 2017. Investigating the extent to azepines). Three key findings emerged regarding changes in
which decreases occurred broadly or were limited to a subset of opioid deaths from July–December 2017 to January–June
opioid types (e.g., prescription opioids versus IMF) and drug com- 2018. First, overall opioid deaths decreased 4.6%. Second,
binations (e.g., IMF co-involving cocaine) can assist in targeting decreases occurred in prescription opioid deaths without co-
of intervention efforts. This report describes opioid deaths during involved illicit opioids and deaths involving non-IMF illicit
January–June 2018 and changes from July–December 2017 in synthetic opioids (fentanyl analogs and U-series drugs) (5).
25¶ of 32 states and the District of Columbia participating in Third, IMF deaths, especially those with multiple illicit opi-
CDC’s State Unintentional Drug Overdose Reporting System oids and common nonopioids, increased. Consequently, IMF
(SUDORS).** Opioid deaths were analyzed by involvement (opi- was involved in approximately two-thirds of opioid deaths
oid determined by medical examiner or coroner to contribute to during January–June 2018. Notably, during January–June
overdose death) of prescription or illicit opioids,†† as well as by the 2018, 62.6% of all opioid deaths co-occurred with at least
§ https://www.cdc.gov/nchs/nvss/vsrr/drug-overdose-data.htm.
one common nonopioid drug. To maintain and accelerate
¶ Alaska, Connecticut, Delaware, Florida, Georgia, Illinois, Kentucky, Maine, reductions in opioid deaths, efforts to prevent IMF-involved
Massachusetts, Minnesota, Missouri, Nevada, New Jersey, New Mexico, North deaths and address polysubstance misuse with opioids must
Carolina, Ohio, Oklahoma, Pennsylvania, Rhode Island, Tennessee, Utah, Vermont, be enhanced. Key interventions include broadening outreach
Virginia, Washington, and Wisconsin. Florida, Illinois, Missouri, Pennsylvania, and
Washington collect data from a subset of counties that accounted for 77%–87% of to groups at high risk for IMF or fentanyl analog exposure
all unintentional or undetermined-intent opioid overdoses in 2017. and overdose. Improving linkage to and engagement in risk-
** SUDORS captures detailed information on toxicology, death scene investigations,
route of administration, and other risk factors that may be associated with a fatal
reduction services and evidence-based treatment for persons
overdose. SUDORS is part of CDC’s Enhanced State Opioid Overdose Surveillance with opioid and other substance use disorders with attention
(ESOOS) program (which funded 12 states through a competitive application to polysubstance use or misuse is also needed.
process in Fiscal Year 2016 and an additional 20 states and the District of Columbia
in Fiscal Year 2017). https://www.cdc.gov/drugoverdose/foa/state-opioid-mm.html.
†† IMF, fentanyl analogs, heroin, and illicitly manufactured U-series drugs.
U-series drugs are novel nonfentanyl-related synthetic opioids with no
authorized medical uses. U-series drug deaths include those involving U-47700
and its analogs U-48800 and U-49900. U-47700, a nonfentanyl benzamide
compound developed by a pharmaceutical company, is not authorized for
medical use in the United States and is currently distributed illicitly for its
heroin-like effect. Deaths involving U-50488 and U-51754 were also included
in this category, but each was involved in five or fewer deaths.

The MMWR series of publications is published by the Center for Surveillance, Epidemiology, and Laboratory Services, Centers for Disease Control and Prevention (CDC),
U.S. Department of Health and Human Services, Atlanta, GA 30329-4027.
Suggested citation: [Author names; first three, then et al., if more than six.] [Report title]. MMWR Morb Mortal Wkly Rep 2019;68:[inclusive page numbers].
Centers for Disease Control and Prevention
Robert R. Redfield, MD, Director
Anne Schuchat, MD, Principal Deputy Director
Chesley L. Richards, MD, MPH, Deputy Director for Public Health Science and Surveillance
Rebecca Bunnell, PhD, MEd, Director, Office of Science
Barbara Ellis, PhD, MS, Acting Director, Office of Science Quality, Office of Science
Michael F. Iademarco, MD, MPH, Director, Center for Surveillance, Epidemiology, and Laboratory Services
MMWR Editorial and Production Staff (Weekly)
Charlotte K. Kent, PhD, MPH, Editor in Chief Martha F. Boyd, Lead Visual Information Specialist
Jacqueline Gindler, MD, Editor Maureen A. Leahy, Julia C. Martinroe,
Mary Dott, MD, MPH, Online Editor Stephen R. Spriggs, Tong Yang,
Terisa F. Rutledge, Managing Editor Visual Information Specialists
Douglas W. Weatherwax, Lead Technical Writer-Editor Quang M. Doan, MBA, Phyllis H. King,
Glenn Damon, Soumya Dunworth, PhD, Teresa M. Hood, MS, Terraye M. Starr, Moua Yang,
Technical Writer-Editors Information Technology Specialists
MMWR Editorial Board
Timothy F. Jones, MD, Chairman
Matthew L. Boulton, MD, MPH Robin Ikeda, MD, MPH Stephen C. Redd, MD
Virginia A. Caine, MD Phyllis Meadows, PhD, MSN, RN Patrick L. Remington, MD, MPH
Katherine Lyon Daniel, PhD Jewel Mullen, MD, MPH, MPA Carlos Roig, MS, MA
Jonathan E. Fielding, MD, MPH, MBA Jeff Niederdeppe, PhD William Schaffner, MD
David W. Fleming, MD Patricia Quinlisk, MD, MPH Morgan Bobb Swanson, BS
William E. Halperin, MD, DrPH, MPH

738 MMWR / August 30, 2019 / Vol. 68 / No. 34 US Department of Health and Human Services/Centers for Disease Control and Prevention
 Morbidity and Mortality Weekly Report

Numbers of opioid deaths of unintentional and undeter- changes in opioid deaths might vary by co-involvement with
mined intent§§ occurring during January–June 2018 and IMF or other illicit opioids, opioid deaths were also grouped
changes from July–December 2017 were analyzed for 25 of into the following eight mutually exclusive categories: 1) IMF
the 32 states and the District of Columbia that participate in with no other illicit opioids involved; 2) IMF co-involving
SUDORS (data for these periods were the most recent and heroin; 3) IMF co-involving fentanyl analogs; 4) co-involved
complete). The states abstract death certificate and medical IMF, heroin, and fentanyl analogs; 5) heroin with no other
examiner and coroner report data, including death scene illicit opioids involved; 6) fentanyl analogs with no other illicit
investigation and toxicology findings. States list drugs involved opioids involved; 7) prescription opioids with no illicit opioids
in (i.e., contributing to) the opioid death as determined by involved; and 8) all other opioid combinations. Finally, deaths
medical examiners and coroners¶¶ and all drugs detected were analyzed by nonopioids (cocaine, methamphetamine, and
(present or co-occurring) by toxicologic tests. Fentanyl and benzodiazepines) that are commonly present and involved in
morphine deaths were classified as prescription opioid deaths opioid deaths.§§§ Tracking the presence of commonly occur-
or illicit opioid deaths based on scene evidence and toxicology ring nonopioids is important to inform public health action
findings.*** Changes in the number of opioid deaths from and has implications for treatment approaches. Some opioid
July–December 2017 to January–June 2018 were analyzed deaths were grouped into one or more of the five opioid type
by five opioid types†††: 1) prescription, 2) IMF, 3) fentanyl categories and nonopioid drug combinations because multiple
analog, 4) heroin, and 5) U-series. Because the frequency and opioids and nonopioids might be involved in a single death
(e.g., an opioid death involving IMF, heroin, cocaine, and a
§§ A death was included in the analysis if 1) the fatal injury occurred within, and
benzodiazepine). Changes in numbers of opioid deaths over
was reported by, one of the 25 SUDORS states, and 2) the death was classified
as an overdose death involving opioids either through review of the medical the analysis period were tested using z-tests or nonoverlapping
examiner/coroner report or the death certificate had International Classification confidence intervals if the number of deaths was <100. SAS
of Diseases, Tenth Revision underlying cause-of-death codes X40–44
(unintentional) or Y10–Y14 (undetermined intent) and multiple cause-of-
statistical software (version 9.4; SAS Institute, Inc.) was used
death codes T40.0, T40.1, T40.2, T40.3, T40.4, or T40.6. Data for this report for all analyses; p-values <0.05 were considered statistically
were downloaded on June 26, 2019, and might differ from earlier or future significant.¶¶¶
reports because states continually update death data and investigations of
suspected drug overdose deaths might involve lengthy investigations. During January–June 2018, among 13,631 opioid deaths in
¶¶ All substances detected were categorized as contributing to the death when the 25 states, data on specific opioids involved were available
deaths were classified by the medical examiner/coroner as a drug overdose for 13,415 (98.4%). IMF was co-involved in 68.0% of 5,281
and toxicology results were available, but no information was available on
the specific drugs contributing to death. heroin deaths and most (82.1%) of 2,678 fentanyl analog
*** Fentanyl was classified on the basis of toxicology, scene, and witness evidence deaths (Table 1). In addition, 1,562 (40.5%) of 3,853 prescrip-
that indicated the likely source as either prescription (e.g., scene evidence of tion opioid deaths co-involved illicit opioids. Opioids com-
fentanyl patches at the overdose location) or illicit (e.g., toxicology evidence
of a fentanyl analog, or scene or witness evidence of illicit drug use including monly involved in opioid deaths were IMF (67.9%), heroin
injection or snorting). If evidence of prescription or illicit use was not (39.4%), prescription opioids (28.7%), and fentanyl analogs
available, fentanyl was categorized as IMF because the vast majority of
fentanyl overdose deaths involve IMF. Morphine in the absence of
(20.0%) (Table 2). Among categories of deaths involving IMF,
6-acetylmorphine (a metabolite of heroin) was classified as likely prescription those with no other illicit opioids involved, those co-involved
morphine (scene or witness evidence of prescription morphine use) or as with heroin, those co-involved with fentanyl analogs, and those
likely heroin (toxicology evidence of heroin impurities or other illicit drug
detected or scene or witness evidence that indicated injection, illicit drug co-involved with heroin and fentanyl analogs accounted for
use, or a history of heroin use). If evidence of prescription or illicit use was
not available, morphine was categorized as prescription because the §§§ Only
investigation did not obtain scene evidence of heroin use or detect commonly occurring (present in >10% of opioid deaths) and
6-acetylmorphine. If morphine was detected along with 6-acetylmorphine, contributing (involved in >50% of opioid deaths in which present)
it was classified as heroin. nonopioids were included. Cutoff for nonopioid inclusion was determined
††† Substances coded as prescription opioids were oxycodone, oxymorphone, by a review of the data.
¶¶¶ To verify that changes in population size did not account for observed
hydrocodone, hydromorphone, tramadol, buprenorphine, methadone,
prescription fentanyl, morphine, codeine, meperidine, tapentadol, changes, sensitivity analyses were conducted in which crude population death
dextrorphan, levorphanol, propoxyphene, noscapine, and pentazocine. Also rates per 100,000 residents for July–December 2017 were compared with
included as prescription drugs were brand names (e.g., Opana) and death rates for January–June 2018 using z-tests or nonoverlapping confidence
metabolites (e.g., nortramadol) of these substances and combinations of intervals if the number of deaths was <100. U.S. Department of Commerce’s
these substances and nonopioids (e.g., acetaminophen-oxycodone). Bureau of Economic Analysis data were used to estimate populations for the
Substances coded as illicit opioids were IMF, heroin, fentanyl analogs, and 25 states in the middle of the last quarter of 2017 and the middle of the
other illicit synthetic opioids, such as U-series drugs (e.g., U-47700), AH- second quarter of 2018 (https://apps.bea.gov/iTable/index.cfm/). For states
7921, and MT-45. This analysis does not distinguish between prescription collecting data on opioid deaths in a subset of counties, the estimates of state
drugs prescribed to the decedent and those that were diverted. Data on population were used and should have introduced minimal bias because of
AH-7921 and MT-45 were not included because they were involved in fewer the >75% coverage of opioid deaths in these states and minimal population
than five deaths. changes over a 6-month period.

US Department of Health and Human Services/Centers for Disease Control and Prevention MMWR / August 30, 2019 / Vol. 68 / No. 34 739
 Morbidity and Mortality Weekly Report

TABLE 1. Number and percentage of opioid overdose deaths that co-involved another opioid, by opioid type (illicitly manufactured fentanyl
[IMF],* fentanyl analogs,† heroin, and prescription opioids§) — 25 states,¶ State Unintentional Drug Overdose Reporting System (SUDORS),
January–June 2018
No. of deaths with co-involved opioid types (%)
Opioid type involved in No. of deaths
opioid death** Jan–Jun 2018 IMF Fentanyl analog Heroin Other illicit opioid†† Prescription opioid
Any suspected IMF 9,105 — 2,199 (24.2) 3,589 (39.4) 4,785 (52.6) 1,250 (13.7)
Any fentanyl analog 2,678 2,199 (82.1) — 1,172 (43.8) 2,366 (88.3) 356 (13.3)
Any suspected heroin 5,281 3,589 (68.0) 1,172 (22.2) — 3,747 (71.0) 796 (15.1)
Any prescription opioid 3,853 1,250 (32.4) 356 (9.2) 796 (20.7) 1,562 (40.5) —
* Among fentanyl-involved deaths, 87.2%, 11.2%, and 1.6% were suspected to involve IMF, had insufficient data to classify the fentanyl death as IMF or prescription
fentanyl, and were suspected to involve prescription fentanyl, respectively. Because the majority of identified cases involved IMF, and characteristics of unclassified
fentanyl deaths were more similar to IMF-involved deaths than to prescription fentanyl–involved deaths, unclassified fentanyl deaths were categorized as suspected
IMF-involved.
† Fentanyl analog-involved deaths included deaths involving carfentanil, acetylfentanyl, acrylfentanyl, furanylfentanyl, 3-methylfentanyl, butyrylfentanyl,
cyclopropylfentanyl, crotonylfentanyl, 4/para-fluorofentanyl, 4/para-fluorobutyrylfentanyl, 4/para-isobutyrylfentanyl, cyclopentylfentanyl, methoxyacetylfentanyl,
isobutyrylfentanyl, furanylethylfentanyl, methoxybutyrlfentanyl, benzylfentanyl, valerylfentanyl, alpha-methylfentanyl, tetrahydrofuranylfentanyl, ocfentanil,
betahydroxythiofentanyl, alfentanil, sufentanil, methylcarfentanil, methylthiofentanyl, phenylfentanyl, omethylacetylfentanyl, and isovalerylfentanyl.
§ Included any opioid deaths involving prescription opioids (oxycodone, oxymorphone, hydrocodone, hydromorphone, tramadol, buprenorphine, methadone,
morphine, codeine, prescription fentanyl, meperidine, tapentadol, dextrorphan, levorphanol, propoxyphene, noscapine, and pentazocine). Other drugs might
have been involved or co-occurred with the prescription opioid.
¶ Alaska, Connecticut, Delaware, Florida, Georgia, Illinois, Kentucky, Maine, Massachusetts, Minnesota, Missouri, Nevada, New Jersey, New Mexico, North Carolina,
Ohio, Oklahoma, Pennsylvania, Rhode Island, Tennessee, Utah, Vermont, Virginia, Washington, and Wisconsin.
** U-series deaths were not reported in the analyses because only 63 deaths involved U-series drugs in the 25 SUDORS states during January–June 2018.
††
Any illicit opioid other than that listed in drug category. For deaths involving illicit opioids (IMF, fentanyl analogs, and heroin) the illicit drug is excluded from this
column. For example, 52.6% of IMF deaths co-involved at least one fentanyl analog, heroin, or U-series drug.

32.2%, 19.1%, 8.7%, and 7.5% of deaths, respectively. Heroin opioids involved (16.6% decline) and fentanyl analog deaths
without other illicit opioids involved accounted for 11.4% of with no other illicit opioids involved (67.9% decline). Declines
deaths, fentanyl analogs with no other illicit opioids involved in heroin deaths with no other illicit opioids involved were
for 2.3%, prescription opioids with no illicit opioids involved offset by increases in heroin deaths co-involving IMF, resulting
for 17.1%, and all other opioid combinations for 1.6%. In the in no significant change in heroin deaths.
Midwest, Northeast, and South U.S. Census regions, deaths The majority of opioid deaths (62.6%) co-occurred with one
involving any IMF were more common than were those involv- or more of the following drugs: benzodiazepines, cocaine, and
ing any heroin. In the West, heroin-involved deaths (47.5%) methamphetamine, which were each present in 32.5%, 34.0%,
were more common than were IMF-involved deaths (15.8%) and 12.1% of deaths, respectively. From July–December
(data not shown). 2017 to January–June 2018, opioid deaths without benzo-
Three principal changes occurred in opioid deaths from diazepines, cocaine, or methamphetamine decreased 8.0%,
July–December 2017 to January–June 2018. First, overall and opioid deaths co-occurring with benzodiazepines signifi-
opioid deaths in the 25 states declined by 4.6% (Table 2). cantly decreased 5.7% (Table 3). Conversely, opioid deaths
Second, declines occurred in prescription opioid deaths with co-occurring with methamphetamine significantly increased
no co-involved illicit opioids (10.6%) and non-IMF illicit by 14.6%. IMF deaths that co-occurred with benzodiazepines,
synthetic opioid deaths, including fentanyl analogs (19.0% cocaine, and methamphetamine significantly increased from
decline) and U-series drugs (75.1% decline). With the excep- July–December 2017 to January–June 2018 by 11.3%, 14.0%,
tion of acetylfentanyl, decreases in fentanyl analog deaths and 31.0%, respectively, as IMF deaths without benzodiaz-
occurred broadly across all fentanyl analogs (52.7% decline). epines, cocaine, or methamphetamine increased 6.7%.
Acetylfentanyl deaths co-involving IMF showed a sharp
Discussion
increase (57.5%). Third, IMF deaths increased by 11.1%
overall, with increases of 9.5%–33.0% in those co-involving Among 25 states participating in SUDORS, three major
other illicit opioids and 9.4% among those with no other illicit changes in opioid deaths from July–December 2017 to
opioids involved. Illicit opioid overdose deaths involving heroin January–June 2018 were identified. These included 1) over-
and fentanyl analogs increased when IMF was co-involved, all decreases in opioid overdose deaths; 2) decreases in both
but decreased when IMF and other illicit opioids were not prescription opioid deaths without co-involved illicit opioids
co-involved. Specifically, increases occurred in IMF deaths co- and non-IMF illicit synthetic opioids (i.e., fentanyl analogs
involving heroin (9.5%), fentanyl analogs (11.4%), and both and U-series drugs) deaths; and 3) increases in IMF deaths,
heroin and fentanyl analogs (33.0%). In contrast, substantial especially those with heroin, fentanyl analogs or nonopioid
declines were observed in heroin deaths with no other illicit drugs. Also, at least one nonopioid drug (benzodiazepines,

740 MMWR / August 30, 2019 / Vol. 68 / No. 34 US Department of Health and Human Services/Centers for Disease Control and Prevention
 Morbidity and Mortality Weekly Report

TABLE 2. Change in the number of opioid overdose deaths, by opioid type, eight common opioid drug combinations, and commonly co-
occurring nonopioids (cocaine, methamphetamine, and benzodiazepines) — 25 states,* State Unintentional Drug Overdose Reporting System
(SUDORS), July–December 2017 to January–June 2018
% Nonopioid drugs commonly present in opioid deaths, Jan–Jun 2018
Opioid deaths with information on Change in no. of opioid deaths,
involved opioids, Jan–Jun 2018, Jul–Dec 2017 to Jan–Jun 2018, Cocaine, Methamphetamine, Benzodiazepines, All three
Characteristic no. (%) no. (%) any† any† any drugs, any
Total opioid overdose deaths 13,415 (100) −648 (−4.6)§ 34.0 12.1 32.5 62.6
% of deaths with contributing NA NA 81.3 81.8 67.5 80.6**
nonopioid drugs present¶
Opioid drug class or drug involved in opioid deaths††
Any prescription opioid§§ 3,853 (28.7) −271 (−6.6)§ 19.8 10.0 50.1 64.3
Any illicit opioid¶¶ 11,124 (82.9) −376 (−3.3)§ 38.6 12.8 28.2 62.6
Any suspected IMF*** 9,105 (67.9) 910 (11.1)§ 39.7 11.2 27.8 62.0
Any suspected heroin 5,281 (39.4) −83 (−1.5) 38.4 13.8 28.5 63.1
Any fentanyl analog ††† 2,678 (20.0) −627 (−19.0)§ 40.3 11.2 30.9 63.6
Any U-series§§§ 63 (0.5) −190 (−75.1)§ 36.5 7.9 39.7 61.9
Common mutually exclusive combinations of opioids involved in opioid deaths¶¶¶
Opioid combinations co-involving IMF
IMF with no other illicit opioids 4,320 (32.2) 370 (9.4)§ 38.3 12.1 27.1 61.7
IMF with heroin 2,566 (19.1) 222 (9.5)§ 40.8 9.2 26.9 61.0
IMF with fentanyl analogs 1,172 (8.7) 120 (11.4)§ 41.6 11.8 29.6 64.2
IMF with heroin and fentanyl 1,008 (7.5) 250 (33.0)§ 40.4 11.8 30.5 63.4
analogs
Illicit opioid combinations not co-involving IMF
Heroin with no other illicit 1,534 (11.4) −306 (−16.6)§ 33.3 23.5 28.9 66.4
opioid
Fentanyl analogs with no other 312 (2.3) −661 (−67.9)§ 35.9 9.6 33.3 61.5
illicit opioid
Prescription opioid with no 2,291 (17.1) −272 (−10.6)§ 11.6 8.7 53.5 62.6
illicit opioid
All other combinations of 212 (1.6) −371 (−63.6)§ 36.3 7.1 36.3 61.8
opioids
Fentanyl analog deaths by acetylfentanyl and IMF co-involvement
Any acetylfentanyl 1,716 (12.8) 590 (52.4)§ 40.9 12.0 29.5 63.6
Acetylfentanyl with IMF 1,685 (12.6) 615 (57.5)§ 41.0 12.0 29.3 63.5
Acetylfentanyl no IMF 31 (0.2) −25 (−44.6) 35.5 9.7 41.9 67.7
All other fentanyl analogs 1,100 (8.2) −1,228 (−52.7)§ 39.8 9.8 32.4 63.5
Other fentanyl analogs with IMF 645 (4.8) −274 (−29.8)§ 42.9 10.9 30.7 64.8
Other fentanyl analogs no IMF 455 (3.4) −954 (−67.7)§ 35.4 8.4 34.7 61.8

Abbreviations: IMF = illicitly manufactured fentanyl; NA = not applicable.

* Alaska, Connecticut, Delaware, Florida, Georgia, Illinois, Kentucky, Maine, Massachusetts, Minnesota, Missouri, Nevada, New Jersey, New Mexico, North Carolina, Ohio, Oklahoma,
Pennsylvania, Rhode Island, Tennessee, Utah, Vermont, Virginia, Washington, and Wisconsin.
† Only the two most frequently co-occurring types of stimulants (cocaine and methamphetamine) are reported because other types of stimulants such as amphetamines did not meet
inclusion criteria.
§ Statistically significantly change from July–December 2017 to January–June 2018 based on z-tests or nonoverlapping confidence intervals if the number of deaths was <100 (p<0.05).
¶ For cocaine, methamphetamine, and benzodiazepines, this row reports a percentage calculated by dividing the number of opioid deaths in which the drug was present and reported
as contributing to the opioid death (numerator) by the number of opioid deaths in which the drug was present (i.e., detected by toxicology tests) irrespective of whether it contributed
to the opioid death (denominator).
** Percentage of all opioid deaths in which cocaine, methamphetamine, or benzodiazepines contributed to death.
†† An opioid death might involve multiple opioids. Thus, total opioid deaths and change in opioid deaths will be different than the sum of the deaths associated with each opioid type.
Other nonopioid drugs might have been involved or co-occurred.
§§ Included any opioid death involving prescription opioids (oxycodone, oxymorphone, hydrocodone, hydromorphone, tramadol, buprenorphine, methadone, morphine, codeine,
prescription fentanyl, meperidine, tapentadol, dextrorphan, levorphanol, propoxyphene, noscapine, and pentazocine). Other drugs might have been involved or co-occurred.
¶¶ Included any opioid death involving IMF, heroin, fentanyl analogs, or U-series drugs. Other drugs might have been involved or co-occurred.
*** Among fentanyl-involved deaths, 87.2%, 11.2%, and 1.6% were suspected to involve IMF, had insufficient data to classify the fentanyl death as IMF or prescription fentanyl, and were
suspected to involve prescription fentanyl, respectively. Because the majority of identified cases involved IMF, and characteristics of unclassified fentanyl deaths were more similar to
IMF-involved deaths than to prescription fentanyl–involved deaths, unclassified fentanyl deaths were categorized as suspected IMF-involved.
††† Fentanyl analog deaths included deaths involving carfentanil, acetylfentanyl, acrylfentanyl, furanylfentanyl, 3-methylfentanyl, butyrylfentanyl, cyclopropylfentanyl, crotonylfentanyl,
4/para-fluorofentanyl, 4/para-fluorobutyrylfentanyl, 4/para-isobutyrylfentanyl, cyclopentylfentanyl, methoxyacetylfentanyl, isobutyrylfentanyl, furanylethylfentanyl, methoxybutyrlfentanyl,
benzylfentanyl, valerylfentanyl, alpha-methylfentanyl, tetrahydrofuranylfentanyl, ocfentanil, betahydroxythiofentanyl, alfentanil, sufentanil, methylcarfentanil, methylthiofentanyl,
phenylfentanyl, omethylacetylfentanyl, and isovalerylfentanyl.
§§§ U-series drugs are novel nonfentanyl-related synthetic opioids with no authorized medical uses. U-series drug deaths include those involving U-47700 and its analogs U-48800 and
U-49900. U-47700, a nonfentanyl benzamide compound developed by a pharmaceutical company, is not authorized for medical use in the United States and is currently distributed
illicitly for its heroin-like effect. Deaths involving U-50488 and U-51754 were also included in this category, but each was involved in five or fewer deaths.
¶¶¶ Six categories are combinations of the illicit opioids involved in death (IMF, heroin, fentanyl analog, and U-series) that were involved in >200 deaths during January–June 2018. These
deaths might co-involve prescription opioids and co-occur with nonopioids. The “prescription opioids with no illicit opioid” category includes only deaths involving prescription opioids
with no illicit opioid co-involvement but might co-occur with other nonopioid drugs. The “all other combinations of opioids” category includes opioid deaths that involved opioid drug
combinations not listed, primarily opioid deaths involving U-series drugs or heroin deaths co-involving fentanyl analogs.

US Department of Health and Human Services/Centers for Disease Control and Prevention MMWR / August 30, 2019 / Vol. 68 / No. 34 741
 Morbidity and Mortality Weekly Report

TABLE 3. Changes in the number and percentage of opioid deaths co-occurring with benzodiazepines, cocaine, and methamphetamine, by
type of opioids involved in death — 25 states,* State Unintentional Drug Overdose Reporting System (SUDORS), July–December 2017 to
January–June 2018
No. of opioid deaths with co-occurring drugs (%)
Type of opioid involved in death Benzodiazepines† Cocaine† Methamphetamine† None of the three drugs
All opioids§ −264 (−5.7)¶ −106 (−2.3) 206 (14.6)¶ −437 (−8.0)¶
Any IMF** 256 (11.3)¶ 445 (14.0)¶ 241 (31.0)¶ 217 (6.7)¶
Illicit opioid, no IMF†† −389 (−39.0)¶ −514 (−42.9)¶ −69 (−14.7)¶ −537 (−43.3)¶
Prescription opioid, no illicit opioid§§ −131 (−9.7)¶ −37 (−12.2) 34 (20.5) −117 (−12.0)¶
Abbreviation: IMF = illicitly manufactured fentanyl.
* Alaska, Connecticut, Delaware, Florida, Georgia, Illinois, Kentucky, Maine, Massachusetts, Minnesota, Missouri, Nevada, New Jersey, New Mexico, North Carolina,
Ohio, Oklahoma, Pennsylvania, Rhode Island, Tennessee, Utah, Vermont, Virginia, Washington, and Wisconsin.
† Opioid deaths co-occurring with benzodiazepines, cocaine and methamphetamine are not mutually exclusive as deaths associated with multiple nonopioids
(e.g., cocaine and benzodiazepines) will be counted in both categories.
§ All opioid deaths (N = 13,415 during January–June 2018).
¶
Statistically significant change from July–December 2017 to January–June 2018 based on z-tests or nonoverlapping confidence intervals if the number of deaths
was <100 (p<0.05).
** All opioid deaths where IMF was involved (N = 9,105 during January–June 2018).
†† All deaths involving other illicit opioids (heroin, fentanyl analogs, and U-series) where IMF was not involved (N = 2,019 during January–June 2018).
§§ Deaths involving prescription opioids without illicit opioid involvement (N = 2,291 deaths during January–June 2018).

cocaine, or methamphetamine) was present in the majority of persons misusing heroin,***** expanding naloxone access,†††††
opioid deaths during January–June 2018. Prescription opioid or changing behaviors of persons injecting drugs to reduce the
deaths stabilized nationally from 2016 to 2017 (6), and the likelihood of an IMF-involved overdose (8).
number of opioid prescriptions filled has been decreasing for Evidence suggests that persons using powdered heroin are
several years,**** as efforts to reduce high-risk prescribing often unaware of whether IMF or fentanyl analogs are pres-
have increased. Findings from this report suggest these efforts ent in illicit products (2,3). Consequently, IMF, heroin, and
might have fostered decreases in prescription opioid deaths fentanyl analog combinations in opioid deaths might represent
without illicit opioids. mixed drug products rather than purposeful co-use. IMF deaths
This report is one of the first to document large decreases without other illicit opioids co-involved and IMF deaths co-
in fentanyl analog and U-series drug deaths across mul- involving heroin are the two most frequent drug combinations
tiple states, including decreases in deaths involving fentanyl in opioid deaths and are consistent with combinations found
analogs (e.g., carfentanil) that drove local outbreaks during when drug products test positive for fentanyl by law enforce-
2016–2017.†††† In contrast, rapid increases in deaths co- ment (2). As IMF supply expanded during January–June,
involving acetylfentanyl and IMF might partly result from 2018 (9), a large, nationally accredited laboratory reported
unintentional production of acetylfentanyl at very low levels that the majority of patients east of the Mississippi River who
during the IMF manufacturing process, rather than deliberate tested positive for heroin also tested positive for fentanyl, and
mixing or co-use of acetylfentanyl with IMF (7). Substantial this percentage increased in early 2018.§§§§§ This suggests
decreases in fentanyl analog and U-series drug deaths even increased mixing of IMF with powdered heroin and fewer
as IMF deaths continue to increase suggest supply changes heroin-only products, consistent with the increases in IMF
requiring investigation. deaths co-involving heroin and decreases in heroin deaths with-
Although concerning, the 6-month 11.1% increase in IMF out IMF documented in this report. In Western states, heroin
deaths in the 25 states is smaller than the approximate dou- deaths predominated, possibly because of the limited mixing of
bling of U.S. fentanyl deaths each year during 2014–2016.§§§§ powdered IMF in black tar heroin, which is distributed primar-
Because IMF is distributed primarily in the powder heroin ily in the West (1,3). Continued vigilance is needed because
market,¶¶¶¶ slower increases in IMF deaths might reflect synthetic opioid (excluding methadone [likely fentanyl]) deaths
successes in one or more objectives: reducing the number of increased in eight states west of the Mississippi in 2017 (6),
persons who initiate heroin use, increasing treatment access for and the IMF supply in the West increased during the first
half of 2018 (9). Finally, increased distribution of counterfeit
**** https://www.cdc.gov/drugoverdose/pdf/pubs/2018-cdc-drug-surveillance-
report.pdf. ***** https://www.hhs.gov/about/news/2018/09/14/samhsa-annual-mental-
†††† https://emergency.cdc.gov/han/han00413.asp.
§§§§ https://www.cdc.gov/nchs/data/nvsr/nvsr67/nvsr67_09-508.pdf. health-substance-use-data-provide-roadmap-for-future-action.html.
††††† https://www.cdc.gov/vitalsigns/naloxone/pdf/vs-0806-naloxone-H.pdf.
¶¶¶¶ https://www.dea.gov/sites/default/files/2018-07/PRB-DIB-003-18.pdf.
§§§§§ https://resource.millenniumhealth.com/signalsreportPR?from=signalsreportPR.

742 MMWR / August 30, 2019 / Vol. 68 / No. 34 US Department of Health and Human Services/Centers for Disease Control and Prevention
 Morbidity and Mortality Weekly Report

prescription pills that contain IMF might increase the risk

Summary
of overdose in persons who use prescription medications not
What is already known about this topic?
prescribed to them, especially opioid pain relievers (2).¶¶¶¶¶
The majority of opioid deaths co-occurred with benzodiaz- Provisional opioid-involved overdose deaths suggest slight
declines from 2017 to 2018, contrasting with sharp increases
epines, cocaine, or methamphetamine highlighting the need during 2014–2017 driven by fentanyl overdose deaths.
to address polysubstance use in the prevention of overdoses
What is added by this report?
and treatment of opioid misuse. Increases in opioid deaths,
From July–December 2017 to January–June 2018 in 25 states,
especially IMF deaths, co-occurring with methamphetamine
opioid deaths decreased 5% overall and decreased for prescrip-
are consistent with previous reports (4) and with increases tion opioids and illicit synthetic opioids excluding illicitly
in methamphetamine supply (9) and methamphetamine use manufactured fentanyl (IMF). However, IMF deaths increased
among persons seeking treatment for opioid misuse (10). 11%. Benzodiazepines, cocaine, or methamphetamine were
Moreover, IMF deaths co-occurring with benzodiazepines and present in 63% of opioid deaths.
cocaine increased during January–June 2018 even as overall What are the implications for public health practice?
opioid deaths co-occurring with benzodiazepines and cocaine Continued increases in IMF deaths highlight the need to
decreased or did not significantly change, respectively. Increases broaden outreach to persons at high risk for IMF overdoses and
in IMF deaths co-occurring with cocaine are consistent with improve linkage to risk-reduction services and evidence-based
treatment. Prevention and treatment efforts should attend to
previous reports (4) and with high co-use of cocaine among
broad polysubstance use/misuse.
persons injecting heroin****** and outbreaks linked to rare
but increasing numbers of drug products that mix IMF and
cocaine (2).†††††† or fentanyl analog exposure and overdose. Improving linkage
The findings in this report are subject to at least five limita- to and engagement in risk-reduction services and evidence-
tions. First, toxicology testing and classification protocols vary based treatment for persons with opioid and other substance
over time and across jurisdictions, which affects whether drugs use disorders with attention to polysubstance use or misuse
were detected and classified as contributing to death. Second, is also needed.
misclassification of prescription and illicit substances might Acknowledgments
occur, but this was minimized by using detailed toxicology
Jurisdictions participating in CDC’s Enhanced State Opioid
results and scene evidence. Third, focus on drugs commonly
Overdose Surveillance (ESOOS) program and providing data in the
involved in opioid deaths might obscure emerging drug issues.
State Unintentional Drug Overdose Reporting System, including
Fourth, patterns in drugs involved in opioid deaths might vary state and jurisdictional health departments, vital registrar offices,
across states and demographic groups. Finally, findings are and coroner and medical examiner offices; the CDC ESOOS team,
limited to the 25 states participating in SUDORS and might Division of Unintentional Injury Prevention, National Center for
not be generalizable to other states. Injury Prevention and Control, CDC; Bruce Goldberger, University
Increases in IMF deaths involving multiple illicit opioids and of Florida College of Medicine, Gainesville, Florida.
benzodiazepines, cocaine, and methamphetamine (nonopioids) Corresponding authors: R. Matt Gladden, mgladden@cdc.gov, 770-488-4276;
highlight the need to better understand how the risk of IMF Puja Seth, pseth@cdc.gov, 404-639-6334.
overdose varies by illicit product potency, variation in potency, 1Division of Unintentional Injury Prevention, National Center for Injury
and form (e.g., powder or counterfeit pill) and a person’s Prevention and Control, CDC.
tolerance or polysubstance use patterns. In response, CDC’s All authors have completed and submitted the International
Overdose Data to Action funding§§§§§§ expands SUDORS Committee of Medical Journal Editors form for disclosure of
from including only opioid-involved deaths to including all potential conflicts of interest. No potential conflicts of interest were
drug overdose deaths to better understand increases in IMF disclosed.
and stimulant and drug combination deaths (with and without
opioids), as well as identify emerging threats. Key interventions References
include broadening outreach to groups at high risk for IMF 1. Zoorob M. Fentanyl shock: The changing geography of overdose
in the United States. Int J Drug Policy 2019;70:40–6. https://doi.
¶¶¶¶¶ https://www.dea.gov/sites/default/files/2018-07/PRB-DIB-003-18.pdf. org/10.1016/j.drugpo.2019.04.010
****** https://www.cdc.gov/hiv/pdf/library/reports/surveillance/cdc-hiv-hssr- 2. US Department of Justice, Drug Enforcement Administration. 2018
nhbs-pwid-2015.pdf. national drug threat assessment. Washington, DC: US Department of
†††††† https://emergency.cdc.gov/han/han00413.asp. Justice, Drug Enforcement Administration; 2018. https://www.dea.
§§§§§§ https://www.cdc.gov/drugoverdose/od2a/index.html. gov/sites/default/files/2018-11/DIR-032-18%202018%20NDTA%20
%5Bfinal%5D%20low%20resolution11-20.pdf

US Department of Health and Human Services/Centers for Disease Control and Prevention MMWR / August 30, 2019 / Vol. 68 / No. 34 743
 Morbidity and Mortality Weekly Report

3. Ciccarone D. The triple wave epidemic: Supply and demand drivers 7. Avedschmidt S, Schmidt C, Isenschmid D, Kesha K, Moons D, Gupta
of the US opioid overdose crisis. Int J Drug Policy 2019;S0955- A. Acetyl fentanyl: trends and concentrations in metro Detroit. J Forensic
3959(19)30018-0. https://doi.org/10.1016/j.drugpo.2019.01.010 Sci 2019;64:149–53. https://doi.org/10.1111/1556-4029.13840
4. Kariisa M, Scholl L, Wilson N, Seth P, Hoots B. Drug overdose deaths 8. Jones CM. Syringe services programs: an examination of legal, policy,
involving cocaine and psychostimulants with abuse potential—United and funding barriers in the midst of the evolving opioid crisis in the
States, 2003–2017. MMWR Morb Mortal Wkly Rep 2019;68:388–95. U.S. Int J Drug Policy 2019;70:22–32. https://doi.org/10.1016/j.
https://doi.org/10.15585/mmwr.mm6817a3 drugpo.2019.04.006
5. Solimini R, Pichini S, Pacifici R, Busardò FP, Giorgetti R. 9. Drug Enforcement Administration, Diversion Control Division. National
Pharmacotoxicology of non-fentanyl derived new synthetic opioids. Front Forensic Laboratory Information System: NFLIS-drug midyear report
Pharmacol 2018;9:654. https://doi.org/10.3389/fphar.2018.00654 2018. Springfield, VA: US Department of Justice, Drug Enforcement
6. Scholl L, Seth P, Kariisa M, Wilson N, Baldwin G. Drug and opioid- Administration; 2019. https://www.nflis.deadiversion.usdoj.gov/
involved overdose deaths—United States, 2013–2017. MMWR Morb DesktopModules/ReportDownloads/Reports/NFLISDrug2018MY.pdf
Mortal Wkly Rep 2018;67:1419–27. https://doi.org/10.15585/mmwr. 10. Ellis MS, Kasper ZA, Cicero TJ. Twin epidemics: the surging rise of
mm675152e1 methamphetamine use in chronic opioid users. Drug Alcohol Depend
2018;193:14–20. https://doi.org/10.1016/j.drugalcdep.2018.08.029

744 MMWR / August 30, 2019 / Vol. 68 / No. 34 US Department of Health and Human Services/Centers for Disease Control and Prevention
 Morbidity and Mortality Weekly Report

Racial Disparities in Breastfeeding Initiation and Duration Among

U.S. Infants Born in 2015
Jennifer L. Beauregard, PhD1,2; Heather C. Hamner, PhD1; Jian Chen, MS1; Wendy Avila-Rodriguez, MPH1;
Laurie D. Elam-Evans, PhD;3 Cria G. Perrine, PhD1

Surveillance of U.S. breastfeeding duration and exclusivity NIS-Child is an ongoing, nationally representative random-
has historically reported estimates among all infants, regardless digit–dialed telephone survey of U.S. households of children
of whether they had initiated breastfeeding. These surveillance aged 19–35 months. From 2011 to 2017, the NIS-Child used
estimates have consistently shown that non-Hispanic black a dual landline and mobile telephone sample frame.§ Although
(black) infants are less likely to breastfeed, compared with other NIS-Child primarily assesses childhood vaccination cover-
racial/ethnic groups.* Less is known about disparities in breast- age, breastfeeding questions were added in 2001 and are the
feeding duration when calculated only among infants who had primary data source for U.S. breastfeeding surveillance. Each
initiated breastfeeding, compared with surveillance estimates cross-sectional survey includes children born in 3 different
based on all infants. CDC analyzed National Immunization calendar years; for this analysis of infants born in 2015, data
Survey-Child (NIS-Child) data for infants born in 2015 to from the 2016–2017 surveys were combined, consistent with
describe breastfeeding duration and exclusivity at ages 3 and national surveillance estimates. Landline sample response rates
6 months among all black and non-Hispanic white (white) were 55.7% in 2016 and 51.9% in 2017. Mobile telephone
infants, and among only those who had initiated breastfeeding. sample response rates were 32.1% in 2016 and 25.0% in
When calculated among all infants regardless of breastfeeding 2017. Children’s breastfeeding history and race/ethnicity were
initiation, breastfeeding differences between black and white reported by their parents or guardians.
infants were 14.7 percentage points (95% confidence interval Breastfeeding initiation rates were calculated for black
[CI] = 10.7–18.8) for any breastfeeding at age 3 months and and white infants born in 2015. Rates of any breastfeeding
were significantly different for both any and exclusive breast- and exclusive breastfeeding (defined as only breast milk and
feeding at both ages 3 and 6 months. Among only infants no solids, water, or other liquids) at ages 3 and 6 months
who had initiated breastfeeding, the magnitude of black-white were calculated for black and white infants using two sets of
differences in breastfeeding rates were smaller. This was most denominators. The first denominator included all infants of
notable in rates of any breastfeeding at 3 months, where the the respective racial/ethnic group regardless of breastfeeding
percentage point difference between black and white infants initiation. The second denominator included only infants of
was reduced to 1.2 (95% CI = -2.3–4.6) percentage points and the respective racial/ethnic group who had initiated breast-
was no longer statistically significant. Black-white disparities feeding. The absolute percentage point difference in each
in breastfeeding duration result, in part, from disparities in breastfeeding rate between black and white infants was also
initiation. Interventions both to improve breastfeeding initia- estimated (hereafter, black-white difference). Estimates were
tion and to support continuation among black mothers might weighted and accounted for the NIS complex sampling design.
help reduce disparities. Data were analyzed using SAS (version 9.4; SAS Institute) and
Breastfeeding has numerous health benefits for infants and SUDAAN (version 11.0.3; RTI International).
mothers. Breastfed infants have reduced risk for ear, respira- Black women were more likely than were white women
tory, and gastrointestinal infections and might be less likely to have incomes <100% of the poverty level (49.3% versus
to develop asthma, obesity, and diabetes (1). Mothers who 17.8%), to receive Special Supplemental Nutrition Program
breastfeed have a lower risk for developing type 2 diabetes, for Women, Infants, and Children benefits (78.2% versus
hypertension, and breast and ovarian cancers (2). U.S. breast- 34.1%), and to be unmarried (65.5% versus 23.9%); they
feeding surveillance has consistently demonstrated that rates also had less education and were younger (Table 1). In 2015,
of breastfeeding initiation, duration, and exclusivity are 10–20 69.4% of black infants initiated breastfeeding, compared with
percentage points lower among black infants, compared with 85.9% of white infants, a difference of 16.5 percentage points
white infants.† (p<0.05) (Table 2).
Among all infants, black infants had a significantly lower rate
* https://www.cdc.gov/breastfeeding/data/nis_data/rates-any-exclusive-bf-socio-
of any breastfeeding at age 3 months (58.0%) than did white
dem-2015.htm.
† https://www.cdc.gov/breastfeeding/data/nis_data/index.htm. § https://www.cdc.gov/breastfeeding/data/nis_data/methods.html.

US Department of Health and Human Services/Centers for Disease Control and Prevention MMWR / August 30, 2019 / Vol. 68 / No. 34 745
 Morbidity and Mortality Weekly Report

TABLE 1. Demographic characteristics of non-Hispanic white and non-Hispanic black infants born in 2015 included in national prevalence estimates
of breastfeeding initiation and duration at ages 3 and 6 months — National Immunization Survey-Child, United States, 2016–2017*
Non-Hispanic white (n = 9,907) Non-Hispanic black (n = 1,607)
Characteristic No. % (95% CI)† No. % (95% CI)†
% of poverty level§
<100 1,312 17.8 (16.5–19.1) 635 49.3 (45.5–53.1)
100–199 1,703 18.7 (17.4–20.0) 366 21.0 (18.2–23.8)
200–399 2,909 27.9 (26.5–29.3) 327 16.1 (13.7–18.4)
400–599 1,967 17.7 (16.5–19.0) 110 5.8 (4.3–7.3)
≥600 2,016 17.9 (16.6–19.3) 169 7.8 (5.2–10.4)
Recipient of WIC¶
Yes 2,723 34.1 (32.5–35.8) 1,137 78.2 (75.5–80.9)
No, but eligible 836 9.0 (8.1–9.8) 107 6.8 (5.0–8.5)
Ineligible 6,298 56.9 (55.2–58.6) 356 15.0 (12.8–17.2)
Mother’s education
Less than high school diploma or GED 460 7.4 (6.3–8.4) 199 16.2 (12.7–19.7)
High school diploma or GED 1,394 20.2 (18.8–21.6) 391 32.2 (28.5–35.8)
Some college 2,435 23.4 (22.0–24.8) 491 26.3 (23.2–29.4)
College graduate 5,618 49.1 (47.4–50.7) 526 25.3 (22.3–28.3)
Mother's age group (yrs)
<20 70 1.1 (0.7–1.5) 40 2.8 (1.6–4.0)
20–29 2,943 34.4 (32.8–36.1) 679 45.2 (41.4–49.0)
≥30 6,894 64.5 (62.8–66.1) 888 52.0 (48.2–55.8)
Mother’s marital status
Married 8,097 76.1 (74.6–77.7) 682 34.5 (31.1–37.8)
Unmarried 1,810 23.9 (22.3–25.4) 925 65.5 (62.2–68.9)
Abbreviations: GED = general educational development certificate; WIC = Special Supplemental Nutrition Program for Women, Infants, and Children.
* Based on National Immunization Survey-Child data from survey years 2016–2017, among infants born in 2015.
† Statistics in this table are based on participants who responded to questions about any breastfeeding at ages 3 and 6 months (N = 11,514). Sample sizes are slightly
smaller for participants who also responded to questions about exclusive breastfeeding at ages 3 and 6 months.
§ Ratio of self-reported family income to the poverty threshold value defined by the U.S. Census Bureau.
¶ Sample sizes for the proportions of participants receiving WIC are slightly smaller due to missing data on WIC status.

TABLE 2. Breastfeeding initiation and duration at ages 3 and 6 months* among non-Hispanic black and non-Hispanic white infants born in
2015 — National Immunization Survey-Child, United States, 2016–2017†
All infants Infants who had initiated breastfeeding
Percentage point Percentage point
Non-Hispanic white Non-Hispanic black difference§ Non-Hispanic white Non-Hispanic black difference§
Breastfeeding
indicator No. % (95% CI) No. % (95% CI) % (95% CI) No. % (95% CI) No. % (95% CI) % (95% CI)
Initiated 9,907 85.9 (84.7 to 87.1) 1,607 69.4 (65.9 to 73.0) 16.5 (12.7 to 20.2)¶ 8,729 N/A 1,159 N/A N/A
breastfeeding
Any 9,907 72.7 (71.2 to 74.2) 1,607 58.0 (54.2 to 61.7) 14.7 (10.7 to 18.8)¶ 8,729 84.7 (83.4 to 85.9) 1,159 83.5 (80.3 to 86.7) 1.2 (−2.3 to 4.6)
breastfeeding
at age 3 mos
Exclusive 9,537 53.0 (51.4 to 54.7) 1,573 36.0 (32.2 to 39.7) 17.0 (12.9 to 21.2)¶ 8,359 62.2 (60.5 to 63.9) 1,125 52.3 (47.8 to 56.9) 9.9 (5.0 to 14.7)¶
breastfeeding
through age
3 mos
Any 9,907 62.0 (60.4 to 63.6) 1,607 44.7 (40.9 to 48.5) 17.3 (13.1 to 21.4)¶ 8,729 72.2 (70.6 to 73.8) 1,159 64.4 (60.2 to 68.6) 7.8 (3.3 to 12.3)¶
breastfeeding
at age 6 mos
Exclusive 9,537 29.5 (28.0 to 31.1) 1,573 17.2 (14.1 to 20.2) 12.4 (8.9 to 15.8)¶ 8,359 34.7 (32.9 to 36.4) 1,125 25.0 (20.8 to 29.2) 9.7 (5.1 to 14.2)¶
breastfeeding
through age
6 mos
Abbreviations: CI = confidence interval; N/A = not applicable.
* Breastfeeding initiation was determined according to participant’s response to the question “Was [child] ever breastfed or fed breast milk?” Breastfeeding duration
was determined according to participant’s response to the question “How old was [child’s name] when [child’s name] completely stopped breastfeeding or being
fed breast milk?” Exclusive breastfeeding was defined as only breast milk (no solids, no water, and no other liquids). To assess the duration of exclusive breastfeeding,
participants were asked two questions about age: 1) “How old was [child’s name] when he/she was first fed formula?” and 2) “How old was [child’s name] when
he/she was first fed anything other than breast milk or formula?” (This includes juice, cow’s milk, sugar water, baby food, or anything else that [child] might have
been given, even water).
† Based on National Immunization Survey-Child data from survey years 2016–2017, among infants born in 2015.
§ Differences in breastfeeding rates between non-Hispanic black and non-Hispanic white infants.
¶ Differences in breastfeeding rates between non-Hispanic black and non-Hispanic white infants are statistically significant (p<0.05, two-sample test of proportions).

746 MMWR / August 30, 2019 / Vol. 68 / No. 34 US Department of Health and Human Services/Centers for Disease Control and Prevention
 Morbidity and Mortality Weekly Report

infants (72.7%); at age 6 months, the rates were 44.7% among among black women (3). Increasing interpersonal support for
black infants and 62.0% among white infants (p<0.05). Rates breastfeeding might help increase breastfeeding initiation and
for exclusive breastfeeding at age 3 months were 36.0% among duration among black women, who might lack breastfeeding
black infants and 53.0% among white infants; at age 6 months, role models in their social networks and be more likely to
the rates were 17.2% among black infants and 29.5% among face negative perceptions of breastfeeding among their peers
white infants (p<0.05) (Table 2). At age 3 months, black-white and communities (3,4). For example, peer counseling might
differences were 14.7 percentage points for any breastfeeding increase breastfeeding initiation and duration among black
(95% CI = 10.7–18.8) and 17.0 percentage points for exclusive mothers (3).
breastfeeding (95% CI = 12.9–21.2). At age 6 months, black- In the United States, the rate of implementation of evidence-
white differences were 17.3 percentage points for any breast- based maternity care practices supportive of breastfeeding is
feeding (95% CI = 13.1–21.4) and 12.4 percentage points for lower among maternity care facilities in neighborhoods with
exclusive breastfeeding (95% CI = 8.9–15.8) (Table 2). larger black populations (5). Hospitals’ use of such practices,
Among only infants who had initiated breastfeeding, the which include helping women initiate breastfeeding within
magnitude of black-white differences in any and exclusive the first hour of birth and not providing breastfeeding infants
breastfeeding rates were smaller (Table 2). This was most with infant formula without a medical indication, increases
notable in rates of any breastfeeding at 3 months, where the rates of breastfeeding initiation, duration, and exclusivity (6).
percentage point difference between black and white infants A recent analysis indicated that making improvements in these
was reduced from 14.7 (95% CI = 10.7–18.8) to 1.2 (95% practices among maternity care facilities in four southern states
CI = -2.3–4.6) percentage points; this difference was no longer reduced black-white disparities in breastfeeding initiation (7).
statistically significant. The black-white difference in exclusive Returning to work is another major barrier to breastfeeding
breastfeeding at age 3 months was reduced from 17.0 percent- initiation and continuation, particularly for black women (3).
age points (95% CI = 12.9–21.2) to 9.9 percentage points A woman’s plans for returning to work are associated with her
(95% CI = 5.0–14.7), in any breastfeeding at 6 months from intention to breastfeed; specifically, women planning to return
17.3 percentage points (95% CI = 13.1–21.4) to 7.8 percentage to work before 12 weeks postpartum, planning to work full-
points (95% CI = 3.3–12.3), and in exclusive breastfeeding at time, or both were less likely to intend to exclusively breast-
age 6 months from 12.4 percentage points (8.9–15.8) to 9.7 feed, compared with women planning to return to work after
percentage points (95% CI = 5.1–14.2). 12 weeks postpartum, planning to work part-time, or both
(8). Black women, especially those with a low income, return
Discussion to work earlier than do women in other racial/ethnic groups
Surveillance of U.S. breastfeeding duration and exclusivity, and are more likely to experience challenges to breastfeeding or
including monitoring for Healthy People 2020¶ objectives, expressing milk, including inflexible work hours (9). Policies
reports estimates among all infants, regardless of whether that enable taking paid leave after giving birth, flexible work
they had initiated breastfeeding. The findings in this report schedules, and support for breastfeeding or expressing milk at
demonstrate that differences between black and white infants work might help improve breastfeeding intention, initiation,
in any and exclusive breastfeeding at ages 3 and 6 months are and duration.**
caused, in part, by racial/ethnic differences in breastfeeding The findings in this report are subject to at least three limi-
initiation. Interventions to improve breastfeeding initiation tations. First, response rates averaged 53.8% for the landline
and support continuation among black mothers might be sample and 28.6% for the mobile telephone sample; further,
important to closing the black-white gap in duration. households without a telephone are not represented. The
Black mothers disproportionately experience a number of possibility exists that selection bias occurs even after adjusting
barriers to breastfeeding, including lack of knowledge about weights for nonresponse and noncoverage. Second, maternal
breastfeeding; lack of peer, family, and social support; insuf- reports of breastfeeding behaviors could be subject to recall bias
ficient education and support from health care settings; and because mothers reported these behaviors when their children
concerns about navigating breastfeeding and employment were aged 19–35 months and to social desirability bias because
(3). Subjective norms, or perceptions of approval from oth- of a desire to provide socially acceptable responses. However,
ers who are important to the person (e.g., family members), maternal recall of breastfeeding behavior has been found to
are important drivers of breastfeeding behaviors, particularly be valid and reliable, especially when recalled within 3 years
(10). Finally, although this report focuses only on black-white
¶ https://www.healthypeople.gov/2020/topics-objectives/topic/maternal-infant-
and-child-health/objectives.
** https://www.cdc.gov/breastfeeding/resources/calltoaction.htm.

US Department of Health and Human Services/Centers for Disease Control and Prevention MMWR / August 30, 2019 / Vol. 68 / No. 34 747
 Morbidity and Mortality Weekly Report

Corresponding author: Jennifer L. Beauregard, uzy2@cdc.gov, 404-498-5337.

Summary
1Divison of Nutrition, Physical Activity, and Obesity, National Center for
What is already known on this topic?
Chronic Disease Prevention and Health Promotion, CDC; 2Epidemic
Rates of breastfeeding duration and exclusivity, calculated for Intelligence Service, CDC; 3Immunization Services Division, National Center
all infants regardless of whether they had initiated breastfeed- for Immunization and Respiratory Diseases, CDC.
ing, are lower among black infants than among white infants. All authors have completed and submitted the International
What is added by this report? Committee of Medical Journal Editors form for disclosure of
Among infants who had initiated breastfeeding, differences potential conflicts of interest. No potential conflicts of interest were
between black infants and white infants in any and exclusive disclosed.
breastfeeding at ages 3 and 6 months were smaller but still
present. References
What are the implications for public health practice? 1. Ip S, Chung M, Raman G, et al. Breastfeeding and maternal and infant
health outcomes in developed countries. Evid Rep Technol Assess (Full
Increasing rates of breastfeeding initiation and supporting
Rep) 2007;153:1–186.
continuation of breastfeeding among black women might help 2. Feltner C, Weber R, Stuebe A, Grodensky C, Orr C, Viswanathan M.
reduce disparities in breastfeeding duration. Strategies might Breastfeeding programs and policies, breastfeeding uptake, and maternal
include improving peer and family support, access to evidence- health outcomes in developed countries. Comparative effectiveness
based maternity care, and employment support. review no. 210. Rockville, MD: US Department of Health and Human
Services, Agency for Healthcare Research and Quality; 2018. https://
www.ncbi.nlm.nih.gov/books/NBK525106
breastfeeding differences, lower rates of breastfeeding duration 3. Jones KM, Power ML, Queenan JT, Schulkin J. Racial and ethnic
and exclusivity among Hispanic infants, compared with non- disparities in breastfeeding. Breastfeed Med 2015;10:186–96. https://
Hispanic white infants, have been documented (3). However, doi.org/10.1089/bfm.2014.0152
4. Louis-Jacques A, Deubel TF, Taylor M, Stuebe AM. Racial and ethnic
because Hispanic and white infants have similar rates of disparities in U.S. breastfeeding and implications for maternal and child
breastfeeding initiation, the methods applied in this report did health outcomes. Semin Perinatol 2017;41:299–307. https://doi.
not affect estimates of breastfeeding duration and exclusivity. org/10.1053/j.semperi.2017.04.007
5. Lind JN, Perrine CG, Li R, Scanlon KS, Grummer-Strawn LM. Racial
Breastfeeding provides optimal nutrition to infants and pro- disparities in access to maternity care practices that support
vides health benefits for both infants and mothers, and CDC breastfeeding—United States, 2011. MMWR Morb Mortal Wkly Rep
works to increase breastfeeding rates among all mothers in the 2014;63:725–8.
6. Pérez-Escamilla R, Martinez JL, Segura-Pérez S. Impact of the baby-
United States. In order to address disparities in breastfeeding friendly hospital initiative on breastfeeding and child health outcomes:
duration, continued efforts are needed to increase rates of a systematic review. Matern Child Nutr 2016;12:402–17. https://doi.
breastfeeding initiation and support continuation of breast- org/10.1111/mcn.12294
feeding among black women. Closing the black-white gap in 7. Merewood A, Bugg K, Burnham L, et al. Addressing racial inequities in
breastfeeding in the southern United States. Pediatrics
breastfeeding duration might require efforts of multiple groups. 2019;143:e20181897. https://doi.org/10.1542/peds.2018-1897
Families, hospitals, and employers can help black women initi- 8. Mirkovic KR, Perrine CG, Scanlon KS, Grummer-Strawn LM. In the
ate and continue breastfeeding, thereby providing their infants United States, a mother’s plans for infant feeding are associated with her
plans for employment. J Hum Lact 2014;30:292–7. https://doi.
with optimal nutrition. org/10.1177/0890334414535665
9. Johnson A, Kirk R, Rosenblum KL, Muzik M. Enhancing breastfeeding
Acknowledgments rates among African American women: a systematic review of current
Katherine Shealy, Division of Nutrition, Physical Activity, and psychosocial interventions. Breastfeed Med 2015;10:45–62. https://doi.
org/10.1089/bfm.2014.0023
Obesity, National Center for Chronic Disease Prevention and Health 10. Li R, Scanlon KS, Serdula MK. The validity and reliability of maternal
Promotion, CDC; Kelley Scanlon, Office of Policy Support, Food and recall of breastfeeding practice. Nutr Rev 2005;63:103–10. https://doi.
Nutrition Service, U.S. Department of Agriculture, Washington, D.C. org/10.1111/j.1753-4887.2005.tb00128.x

748 MMWR / August 30, 2019 / Vol. 68 / No. 34 US Department of Health and Human Services/Centers for Disease Control and Prevention
 Morbidity and Mortality Weekly Report

Notes from the Field

Mumps in Detention Facilities that House Reference Centers, or commercial laboratories. Sequencing
Detained Migrants — United States, September of isolates from 70 patients identified genotype G, the most
2018–August 2019 common mumps genotype detected in the United States since
Jessica Leung, MPH1; Diana Elson, DrPH2; Kelsey Sanders, MPH3; 2006 (2). IHSC provided >25,000 doses of measles-mumps-
Mona Marin, MD1; Greg Leos, MPH3; Brandy Cloud, DNP2; Rebecca J. rubella (MMR) vaccine in response to mumps in 56 facilities.
McNall, PhD1; Carole J. Hickman, PhD1; Mariel Marlow, PhD1 Since 2015, approximately 150 mumps outbreaks and
16,000 cases have been reported in the United States, typi-
On October 12, 2018, five confirmed cases of mumps among
cally in close-contact settings such as universities, schools, and
migrants who had been transferred between two detention
athletic events.§ This is the first report of mumps outbreaks
facilities were reported by the facilities to the Texas Department
in detention facilities.
of State Health Services (TDSHS). By December 11, eight
MMR vaccination efforts differ among detention facilities;
Texas detention facilities and six facilities in five other states had
facilities should follow local or state health department rec-
reported 67 mumps cases to U.S. Immigration and Customs
ommendations for preventing and responding to mumps (3)
Enforcement (ICE) Health Service Corps (IHSC) or local
and should report cases and follow disease control guidance
health departments. On December 12, TDSHS contacted
from their health department. Detainees and staff members at
CDC to discuss mumps control in detention facilities and
increased risk for mumps should be offered MMR vaccine per
facilitate communication with IHSC. During January 4–17,
existing recommendations for vaccination during outbreaks
2019, six more state health departments reported new cases
(4,5). MMR vaccine has not been shown to be effective at
in detention facilities, which prompted CDC and IHSC to
preventing disease in persons already infected with mumps;
launch a coordinated national outbreak response.
facilities should be aware that cases might occur among detain-
During September 1, 2018–August 22, 2019, a total of 898
ees exposed before vaccination.
confirmed and probable mumps cases (1) in adult migrants
Health departments, CDC, IHSC, and facility health admin-
detained in 57 facilities (18% of 315 U.S. facilities that
istration can work together to develop appropriate control
house ICE detainees*) were reported in 19 states (Figure); an
measures based on local epidemiology and the specific needs
additional 33 cases occurred among staff members. Private
of each facility. Identifying and vaccinating close contacts of
companies operated 34 facilities, 19 were county jails that
exposed or symptomatic persons with mumps in detention
house detained migrants, and four were ICE-operated. Forty-
centers is challenging. IHSC can look up transfer history and
four percent (394) of cases were reported from facilities that
facilitate vaccine procurement for detainees in ICE custody
house ICE detainees in Texas. Median patient age was 25 years
upon request from facility health services administrators. CDC
(range = 17–67); 846 (94%) were male. Based on detainee
is coordinating communication among state and local health
custody status during their incubation period (12–25 days
departments, IHSC, and other federal partners to mobilize
before symptom onset), most (758, 84%) patients were exposed
appropriate resources and is providing technical support
while in custody of ICE or another U.S. agency†; 43 (5%) were
for implementing appropriate disease control and preven-
exposed before apprehension; and the custody status at the time
tion measures. Effective public health interventions require
of exposure of 97 (11%) was unknown. Among those with data
understanding of facility and custody operations, which often
on complications, 79 (15%) of 527 male patients reported
involve frequent transfers of detainees (between facilities and
orchitis, and at least 13 patients were hospitalized. More than
states) and multiple entities with authority for operations and
half (576, 64%) of cases were confirmed by quantitative reverse
detainee custody.
transcription–polymerase chain reaction testing or viral culture
As of August 22, 2019, mumps outbreaks are ongoing in 15
testing at CDC, state public health laboratories, Association of
facilities in seven states, and new introductions into detention
Public Health Laboratories–CDC Vaccine Preventable Disease
facilities through detainees who are transferred or exposed
before being taken into custody continue to occur.
* Personal communication, Dr. Diana Elson, U.S. Immigration and Customs
Enforcement. Facility count of 315 detention facilities on August 13, 2019,
§ https://www.cdc.gov/mumps/outbreaks.html.
that housed ICE detainees with an average daily population >0 during Fiscal
Year 2019. The number of facilities might change with time.
† U.S. Customs and Border Protection and U.S. Marshals Service.

US Department of Health and Human Services/Centers for Disease Control and Prevention MMWR / August 30, 2019 / Vol. 68 / No. 34 749
 Morbidity and Mortality Weekly Report

FIGURE. Mumps cases among U.S. Immigration and Customs Enforcement (ICE) detainees, by custody status* at time of exposure, by week of
onset — United States, September 2018–August 2019 (N = 898)†
45

40 Unknown
In ICE/CBP/USMS custody
35 Before apprehension

30
No. of cases

25

20

15

10

0
36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
2018 2019
Epidemiologic week and year

Abbreviations: CBP = U.S. Customs and Border Protection; USMS = U.S. Marshals Service.
* Based on mumps incubation period of 12–25 days before symptom onset.
† Data collected as of August 22, 2019.

Acknowledgments All authors have completed and submitted the International

State and local health departments; Nakia Clemmons, Gimin Committee of Medical Journal Editors form for disclosure of
Kim, Don Latner, Adria Lee, Manisha Patel, Holly Patrick, Sarah potential conflicts of interest. No potential conflicts of interest were
Poser, Paul Rota, Jeanette St. Pierre, Adam Wharton, Division of disclosed.
Viral Diseases, National Center for Immunization and Respiratory References
Diseases, CDC; Nathan Crawford, Immunization Services Division,
1. Council of State and Territorial Epidemiologists. CSTE position
National Center for Immunization and Respiratory Diseases, CDC; statement, 11-ID-18: mumps 2012 case definition. Atlanta, GA: Council
Amanda Cohn, Mary Ann Hall, Office of the Director, National of State and Territorial Epidemiologists; 2012. https://wwwn.cdc.gov/
Center for Immunization and Respiratory Diseases, CDC; Nasim nndss/conditions/mumps/case-definition/2012/
Farach, Division of Global HIV and TB, Deputy Director for 2. Rubin SA, Qi L, Audet SA, et al. Antibody induced by immunization
Public Health Service and Implementation Science, CDC; Dakota with the Jeryl Lynn mumps vaccine strain effectively neutralizes a
heterologous wild-type mumps virus associated with a large outbreak.
McMurray, Geri Tagliaferri, Public Health Safety and Preparedness, J Infect Dis 2008;198:508–15. https://doi.org/10.1086/590115
DHS/ICE/ERO/ICE Health Service Corps; Stephanie Daniels, 3. Clemmons N, Hickman C, Lee A, Marin M, Patel M. Manual for the
Jeff Haug, Jim Hicks, Jai Patel, Clinical Services Unit, Pharmacy surveillance of vaccine-preventable diseases. Chapter 9: mumps. Atlanta,
Program DHS/ICE/ERO/ICE Health Service Corps; Association GA: US Department of Health and Human Services, CDC; 2018. https://
of Public Health Laboratories–CDC Vaccine Preventable Disease www.cdc.gov/vaccines/pubs/surv-manual/chpt09-mumps.html
4. CDC. CDC guidance for public health authorities on use of a 3rd dose
Reference Centers: California Department of Public Health, of MMR vaccine during mumps outbreaks. Atlanta, GA: US Department
Richmond, California; Minnesota Department of Health, St. Paul, of Health and Human Services, CDC; 2019. https://www.cdc.gov/
Minnesota; New York State Department of Health, Wadsworth mumps/health-departments/MMR3.html
Center, Albany, New York; Wisconsin State Laboratory of Hygiene, 5. Marin M, Marlow M, Moore KL, Patel M. Recommendation of the
Madison, Wisconsin. Advisory Committee on Immunization Practices for use of a third dose
of mumps virus–containing vaccine in persons at increased risk for mumps
Corresponding author: Jessica Leung, ncirddvdmmrhp@cdc.gov. during an outbreak. MMWR Morb Mortal Wkly Rep 2018;67:33–8.
1Division of Viral Diseases, National Center for Immunization and Respiratory https://doi.org/10.15585/mmwr.mm6701a7
Diseases, CDC; 2Public Health Safety and Preparedness Unit, DHS/ICE/ERO/
ICE Health Service Corps; 3Emerging and Acute Infectious Disease Branch,
Texas Department of State Health Services.

750 MMWR / August 30, 2019 / Vol. 68 / No. 34 US Department of Health and Human Services/Centers for Disease Control and Prevention
 Morbidity and Mortality Weekly Report

Notes from the Field

Multistate Outbreak of Salmonella Agbeni SNP) to the isolate cultured from the Oregon white cake mix.
Associated with Consumption of Raw Cake On October 25, CDC requested officials in Maryland, Ohio,
Mix — Five States, 2018 and Wisconsin to interview patients using a questionnaire with
Stephen G. Ladd-Wilson, MS1; Karim Morey, MS2; Sarah E. Koske, specific questions about baking exposures.
DVM3; Bailey Burkhalter, MPH4; Lyndsay Bottichio, MPH5; Joshua On October 31, the Food and Drug Administration (FDA)
Brandenburg5; John Fontana, PhD2; Kristina Tenney6; Kirthi K. initiated an investigation of manufacturer A with regard to
Kutumbaka, PhD6; Mansour Samadpour, PhD6; Katherine Kreil, MPH7;
Paul R. Cieslak, MD1
the Salmonella-positive white cake mix. In addition to the
investigation and document collection, FDA collected samples
In August 2018, two Oregon patients with diagnosed including an ingredient (flour), finished cake mix, and envi-
Salmonella infection were interviewed using a standard enteric ronmental samples. All collected samples tested negative for
illness questionnaire; both patients reported having eaten raw Salmonella. On November 5, a voluntary recall of manufacturer
cake mix. Standardized interview questionnaire data collected A’s classic white, classic butter golden, signature confetti, and
from 207 Oregon patients with salmonellosis in 2017 indicated classic yellow cake mixes was announced because they might
a 5% rate of consumption of raw “cake mix or cornbread mix” be contaminated with Salmonella bacteria.
(Oregon Health Authority, unpublished data, 2017). The bino- On January 14, 2019, CDC declared this outbreak, which
mial probability that both 2018 patients were exposed to raw totaled seven cases in five states,† to be over (1). This is the first
cake mix by chance was determined to be 0.003, prompting time that OHA used WGS data on the publicly available NCBI
the Oregon Health Authority (OHA) to collect and test the website to detect a multistate outbreak associated with a widely
contents of 43 boxes of unopened cake mix of various brands distributed consumer product, which resulted in product action.
from six retail locations. OHA sent samples to the Institute WGS of food and environmental isolates and subsequent analysis
for Environmental Health Laboratories in Lake Forest Park, on the NCBI and SEDRIC platforms are emerging as useful
Washington, for pathogen testing. Salmonella Agbeni was tools in identifying outbreaks associated with widely distributed
isolated from an unopened box of white cake mix from manu- products with long shelf lives and low background rates of con-
facturer A, and whole genome sequencing (WGS) data describ- sumption, such as raw cake mix. Detection of these outbreaks is
ing the isolate were uploaded to the U.S. National Library of typically difficult and relies mainly upon epidemiologic evidence
Medicine’s National Center for Biotechnology Information from investigation of a larger number of cases (2–4). These
(NCBI) website (https://www.ncbi.nlm.nih.gov/pathogens). efforts also highlight the value of collaboration between public
OHA used the NCBI database to compare sequence data with health epidemiologists and laboratorians as well as the use of new
the cake mix isolate (PNUSAS056022) and then consulted technological tools for outbreak detection. During outbreak or
CDC’s System for Enteric Disease Response, Investigation, cluster investigations, food and environmental samples should
and Coordination (SEDRIC), a web-based, outbreak investiga- be collected as quickly as possible whenever practical, particu-
tion tool designed for collaborative, multistate investigations larly when epidemiologic data suggest an association. WGS, in
of enteric disease outbreaks.* On October 19, OHA deter- conjunction with the NCBI website and SEDRIC, can be used
mined that clinical isolates from four patients from Maryland, to identify genetically related isolates quickly.
Ohio, and Wisconsin, with specimen isolation dates ranging † Florida, Maryland, Missouri, Ohio, and Wisconsin.
from June to September 2018, were genetically related to the
Salmonella Agbeni isolate from the unopened box of white cake Corresponding author: Stephen Ladd-Wilson, Stephen.G.Ladd-Wilson@state.or.us,
mix, within four single nucleotide polymorphisms (SNPs). 971-673-0138.
On October 22, 2018, OHA notified state public health 1Oregon Health Authority, Public Health Division; 2Oregon Health Authority,
counterparts in the three states of this finding and inquired Oregon State Public Health Laboratory; 3Wisconsin Division of Public Health;
4Douglas County Public Health, Roseburg, Oregon; 5National Center for
about raw cake mix exposures among their patients. The
Emerging Zoonotic Infectious Diseases, CDC; 6Institute for Environmental
Wisconsin patient reported having consumed an entire box Health Laboratories, Lake Forest Park, Washington; 7Coordinated Outbreak
of raw white cake mix over several days during the likely Response and Evaluation Network, Food and Drug Administration, Silver
exposure period. In addition, WGS analysis indicated that Spring, Maryland.
this clinical isolate was closely related genetically (within one All authors have completed and submitted the International
Committee of Medical Journal Editors form for disclosure of potential
* https://www.cdc.gov/foodsafety/outbreaks/investigating-outbreaks/sedric.html. conflicts of interest. No potential conflicts of interest were disclosed.

US Department of Health and Human Services/Centers for Disease Control and Prevention MMWR / August 30, 2019 / Vol. 68 / No. 34 751
 Morbidity and Mortality Weekly Report

References 3. Crowe SJ, Bottichio L, Shade LN, et al. Shiga toxin–producing E. coli
infections associated with flour. N Engl J Med 2017;377:2036–43. https://
1. CDC. Outbreak of Salmonella infections: investigation notice. Atlanta,
doi.org/10.1056/NEJMoa1615910
GA: US Department of Health and Human Services, CDC; 2019. https://
4. McCallum L, Paine S, Sexton K, et al. An outbreak of Salmonella
www.cdc.gov/salmonella/agbeni-11-18/
Typhimurium phage type 42 associated with the consumption of raw
2. Morton V, Cheng JM, Sharma D, Kearney A. Notes from the field: an
flour. Foodborne Pathog Dis 2013;10:159–64. https://doi.org/10.1089/
outbreak of Shiga toxin–producing Escherichia coli O121 infections
fpd.2012.1282
associated with flour—Canada, 2016–2017. MMWR Morb Mortal Wkly
Rep 2017;66:705–6. https://doi.org/10.15585/mmwr.mm6626a6

752 MMWR / August 30, 2019 / Vol. 68 / No. 34 US Department of Health and Human Services/Centers for Disease Control and Prevention
 Morbidity and Mortality Weekly Report

QuickStats

FROM THE NATIONAL CENTER FOR HEALTH STATISTICS

Percentage* of Currently Employed Adults Who Have Paid Sick Leave,† by

Industry§ — National Health Interview Survey, 2009 and 2018¶

All industries
2009
2018
Agriculture, forestry, & fishing

Construction

Wholesale & retail trade

Services
Industry

Mining

Manufacturing

Transportation, communications,
& electricity

Finance, insurance, & real estate

Public administration

0 20 40 60 80 100
Percentage

* With 95% confidence intervals shown with error bars.

† Based on responses to a question that asked, “Do you have paid sick leave on this main job or business?”
§ Respondents were asked to identify the business or industry of their main job, and these industries/businesses
were then categorized by the North American Industry Classification System (https://www.census.gov/eos/
www/naics).
¶ Estimates were based on a sample of the U.S. civilian, noninstitutionalized population aged ≥18 years. Adults not
currently employed at the time of interview were not included in the denominators when calculating percentages.

The percentage of all currently employed workers with access to paid sick leave increased from 57.8% in 2009 to 62.4% in 2018. By
industry, the percentage increased for workers in construction (32.7% to 43.9%), wholesale & retail trade (48.3% to 53.1%), services
(56.7% to 60.8%), and manufacturing (60.7% to 65.5%). In 2018, fewer than half of workers in agriculture, forestry, and fishing
and construction industries had access to paid sick leave compared to approximately 90% of workers in public administration.
Source: National Health Interview Survey, 2009–2018. https://www.cdc.gov/nchs/nhis.htm.
Reported by: Abay Asfaw, PhD, AAsfaw@cdc.gov, 202-245-0635; Roger R. Rosa, PhD; Regina Pana-Cryan, PhD.

US Department of Health and Human Services/Centers for Disease Control and Prevention MMWR / August 30, 2019 / Vol. 68 / No. 34 753
 Morbidity and Mortality Weekly Report

The Morbidity and Mortality Weekly Report (MMWR) Series is prepared by the Centers for Disease Control and Prevention (CDC) and is available free
of charge in electronic format. To receive an electronic copy each week, visit MMWR at https://www.cdc.gov/mmwr/index.html.
Readers who have difficulty accessing this PDF file may access the HTML file at https://www.cdc.gov/mmwr/index2019.html. Address all inquiries about the
MMWR Series, including material to be considered for publication, to Executive Editor, MMWR Series, Mailstop E-90, CDC, 1600 Clifton Rd., N.E.,
Atlanta, GA 30329-4027 or to mmwrq@cdc.gov.
All material in the MMWR Series is in the public domain and may be used and reprinted without permission; citation as to source, however, is appreciated.
MMWR and Morbidity and Mortality Weekly Report are service marks of the U.S. Department of Health and Human Services.
Use of trade names and commercial sources is for identification only and does not imply endorsement by the U.S. Department of Health and Human Services.
References to non-CDC sites on the Internet are provided as a service to MMWR readers and do not constitute or imply endorsement of these organizations
or their programs by CDC or the U.S. Department of Health and Human Services. CDC is not responsible for the content of these sites. URL addresses
listed in MMWR were current as of the date of publication.

ISSN: 0149-2195 (Print)