The Changing Epidemiology of Candidemia in the United States: Injection Drug Use as an Increasingly Common Risk Factor—Active Surveillance in Selected Sites, United States, 2014–2017

Abstract

Background

Injection drug use (IDU) is a known, but infrequent risk factor on candidemia; however, the opioid epidemic and increases in IDU may be changing the epidemiology of candidemia.

Methods

Active population-based surveillance for candidemia was conducted in selected US counties. Cases of candidemia were categorized as IDU cases if IDU was indicated in the medical records in the 12 months prior to the date of initial culture.

Results

During 2017, 1191 candidemia cases were identified in patients aged >12 years (incidence: 6.9 per 100 000 population); 128 (10.7%) had IDU history, and this proportion was especially high (34.6%) in patients with candidemia aged 19–44. Patients with candidemia and IDU history were younger than those without (median age, 35 vs 63 years; P < .001). Candidemia cases involving recent IDU were less likely to have typical risk factors including malignancy (7.0% vs 29.4%; relative risk [RR], 0.2 [95% confidence interval {CI}, .1–.5]), abdominal surgery (3.9% vs 17.5%; RR, 0.2 [95% CI, .09–.5]), and total parenteral nutrition (3.9% vs 22.5%; RR, 0.2 [95% CI, .07–.4]). Candidemia cases with IDU occurred more commonly in smokers (68.8% vs 18.5%; RR, 3.7 [95% CI, 3.1–4.4]), those with hepatitis C (54.7% vs 6.4%; RR, 8.5 [95% CI, 6.5–11.3]), and in people who were homeless (13.3% vs 0.8%; RR, 15.7 [95% CI, 7.1–34.5]).

Conclusions

Clinicians should consider injection drug use as a risk factor in patients with candidemia who lack typical candidemia risk factors, especially in those with who are 19–44 years of age and have community-associated candidemia.

candidemia, fungal infections, injection drug use

(See the Editorial Commentary by Andes on pages 1738–40.)

Candidemia is one of the most common healthcare-associated bloodstream infections and generally occurs days to weeks into a hospitalization [1, 2]. These infections result in increased length of stay, high medical costs, and crude mortality of up to 30% [3]. Established risk factors among adults include prolonged hospitalization [4], intensive care unit (ICU) admission, recent abdominal surgery, neutropenia, organ transplantation, solid organ malignancy, hemodialysis, recent use of broad-spectrum antibiotics, total parenteral nutrition, and presence of central venous catheters (CVCs) or indwelling devices [5, 6]. Both noninvasive and invasive Candida infections have been linked to injection drug use (IDU) [7, 8], especially community-onset candidemia, but the extent to which IDU contributes to candidemia is not known.

The Centers for Disease Control and Prevention (CDC) conducts active, population-based surveillance for candidemia in 9 metropolitan areas across the United States, collecting detailed medical information on all cases detected within defined surveillance areas. In light of the opioid crisis, increasing proportion of heroin use and disorder [9] and rates of IDU [10], and reports of increases in other acute infections such as invasive methicillin-resistant Staphylococcus aureus (MRSA) associated with IDU [11], we aimed to assess whether IDU may be affecting the epidemiology of candidemia infections in the United States.

METHODS

Surveillance Population

Active, population-based surveillance for candidemia was conducted in 45 counties in 9 states (California, Colorado, Georgia, Maryland, Minnesota, New Mexico, New York, Oregon, Tennessee) with a combined population of nearly 18 million persons during 2017. A total of 167 hospitals and 97 laboratories serve these catchment areas. Data available from 2014 to 2017 from 21 counties in 4 sites (Georgia, Maryland, Oregon, Tennessee) were used to examine trends. A site refers to counties under surveillance within a given state.

Surveillance Methods

Clinical, reference, and commercial laboratories that serve the population in the surveillance catchment areas were recruited to participate in the surveillance program. Laboratories reported blood cultures positive for Candida spp to the local surveillance officer. A periodic audit of the laboratory records was conducted to ensure completeness of reporting. A case was defined as a blood culture with Candida spp in a surveillance-area resident. Any other blood culture positive for Candida spp within 30 days after the incident blood culture in the same patient was considered part of the same case. A patient could have multiple cases of candidemia if they had blood cultures with Candida spp collected >30 days from the incident blood culture. Positive Candida spp blood cultures from residents outside the surveillance counties were excluded.

If the case criteria were met, trained surveillance officers reviewed medical records and used a standardized case-report form to collect basic demographic and clinical information, including any documentation of IDU in the past 12 months. Cases with mention of IDU in the past 12 months in the medical chart are referred to as IDU cases and the remaining as non-IDU cases. Cases in children under the age of 12 were excluded from the dataset, as children under that age are unlikely to inject drugs. Cases with positive blood cultures obtained after 3 days of hospital admission (with admission being day 1) were categorized as healthcare associated. Cases with positive blood cultures obtained in the first 3 days of admission with recent healthcare exposure (previously defined by Cleveland et al, 2015 [12]) were categorized as healthcare associated, community onset. Cases with positive blood cultures obtained in the first 3 days of hospital admission without recent healthcare exposure were categorized as community associated.

Statistical Analysis

Demographic variables were analyzed at the patient level. All other variables were analyzed at the case level since the measures (such as underlying conditions, ICU admission, receipt of antibiotics, etc) could change between each case of candidemia that occur in the same patient. Categorical variables were analyzed using χ ² or Fisher exact tests. Continuous variables were analyzed using Wilcoxon rank-sum test (median). Unadjusted relative risk (RR) was calculated to assess the association between patient characteristics and IDU status. Candidemia cases without history of IDU use was used as the reference group. Test for trend was conducted using Cochran-Armitage trend test. The 2-tailed level of significance was set at α = .05. All analyses were performed in SAS software (version 9.4, SAS Institute, Cary, North Carolina).

Human Subjects

The CDC Human Subjects Office determined that this was a nonresearch activity. Institutional review boards (IRB) at all 9 sites individually reviewed this project and deemed it either nonresearch public health surveillance or human subjects research, and in those cases it was approved by the IRB.

RESULTS

During 2017, we identified 1191 cases of candidemia among 1139 patients in the 9 surveillance sites, representing an overall incidence of 6.9 per 100 000 population. Thirty cases under the age of 12 were excluded from the analysis.

Of the 1191 cases in persons aged >12 years, 128 (10.7%) had IDU recorded in their medical charts. The proportion of cases with IDU history ranged from 1.4% in Minnesota to 32.3% in New Mexico (Figure 1). In the 4 sites with pre-2017 data, the proportion of cases with IDU history increased from 6.9% in 2014 to 15.2% in 2017 (P = .001), with the proportion in Tennessee nearly tripling from 6.5% to 18.2% during the same period.

Figure 1.

Proportion of candidemia cases with injection drug use (IDU) history by EIP surveillance site, 2017 (n = 1191). Abbreviations: CA, California; CO, Colorado; EIP, Emerging Infections Program; GA, Georgia; MD, Maryland; MN, Minnesota; NM, New Mexico; NY, New York, OR, Oregon; TN, Tennessee.

Patients with candidemia and IDU history were significantly younger than patients with candidemia and no IDU history (median age, 35 vs 63 years; P < .001; Table 1). Among patients with candidemia who were 19–44 years old, 34.6% had IDU history; 8.2% of those who were 45–64 years and 0.4% of those ≥65 years had IDU history. Patients with IDU history were significantly less likely than those without IDU history to be black (15.5% vs 30.3%; RR, 0.5 [95% confidence interval {CI}, .3–.8]).

Table 1.

Patient-level Demographic Characteristics and Clinical Characteristics of Injection Drug Use (IDU) and Non-IDU Candidemia Cases Among Emerging Infections Program Candidemia Sites, 2017

Characteristic	Injection Drug Use	No Injection Drug Use	Total	RR (95% CI)b	P Value
Patient demographicsa	n = 116	n = 996	N = 1112
Age, y, median (IQR)	35 (30–49)	63 (52–73)	61 (47–71)	…	<.001c
Sex
Male	63 (54.3)	550 (55.2)	613 (55.1)	0.98 (.82–1.17)
Race
White	91 (78.5)	581 (58.3)	672 (60.4)	1.3 (1.2–1.5)
Black	18 (15.5)	302 (30.3)	320 (28.8)	0.5 (.3–.8)
American Indian/Alaska Native	1 (0.9)	6 (0.6)	7 (0.6)	1.4 (.2–11.8)
Asian	0 (0)	24 (2.4)	24 (2.2)	NA
Native Hawaiian/Pacific Islander	0 (0)	6 (0.6)	6 (0.5)	NA
Multiracial	0 (0)	1 (0.1)	1 (0.1)	NA
Unknown	6 (5.2)	76 (7.6)	82 (7.4)	0.7 (.3–1.5)
Ethnicity
Non-Hispanic	99 (85.3)	824 (82.7)	923 (83.0)	1.0 (1.0–1.1)
Hispanic	3 (2.6)	62 (6.2)	65 (5.9)	0.4 (.1–1.3)
Underlying conditionsd	n = 128	n = 1063	n = 1191
Alcohol abuse, current	10 (7.8)	75 (7.1)	85 (7.1)	1.1 (.6–2.1)
Chronic cognitive deficit/dementia	2 (1.6)	79 (7.4)	81 (6.8)	0.2 (.05–.8)
Chronic kidney disease	7 (5.5)	315 (29.7)	322 (27.1)	0.2 (.09–.4)
Chronic liver disease	73 (57.0)	149 (14.0)	222 (18.7)	4.1 (3.3–5.0)
Cirrhosis	9 (7.0)	85 (8.0)	94 (7.9)	0.9 (.5–1.7)
Ascites	2 (1.6)	34 (3.2)	36 (3.0)	0.5 (.1–2.0)
Hepatic encephalopathy	1 (0.8)	28 (2.6)	29 (2.4)	0.3 (.04–2.2)
Hepatitis C	70 (54.7)	68 (6.4)	138 (11.6)	8.5 (6.5–11.3)
Chronic pulmonary disease	8 (6.3)	239 (22.5)	246 (30.7)	0.2 (.1–.5)
Congestive heart failure	9 (7.0)	176 (16.6)	185 (15.6)	0.4 (.2–.8)
Stroke/TIA	2 (1.6)	155 (14.6)	157 (13.2)	0.1 (.03–.4)
Diabetes	19 (14.8)	374 (35.2)	393 (33.0)	0.4 (.3–.6)
HIV	7 (5.5)	21 (2.0)	28 (2.4)	2.8 (1.2–6.4)
Malignancy	9 (7.0)	312 (29.4)	321 (27.0)	0.2 (.1–.5)
Hematologic	4 (3.1)	55 (5.2)	59 (5.0)	0.6 (.2–1.6)
Solid organ (nonmetastatic)	5 (3.9)	165 (15.5)	470 (39.5)	0.3 (.1–.6)
Neurological condition	6 (4.7)	114 (10.7)	120 (10.1)	0.4 (.2–1.0)
Obesity or morbid obesity	7 (5.5)	126 (11.9)	133 (11.2)	0.5 (.2–1.0)
Smoker, current	88 (68.8)	196 (18.5)	284 (23.9)	3.7 (3.1–4.4)
Pancreatitis	4 (3.1)	46 (4.3)	50 (4.2)	0.7 (.3–2.0)
Exposures in the healthcare setting	n = 128	n = 1063	N = 1191
Any surgery	32 (25.0)	374 (35.2)	406 (34.2)	0.7 (.5–1.0)
Abdominal surgery	5 (3.9)	189 (17.5)	195 (16.2)	0.2 (.09–.5)
Total parenteral nutrition	5 (3.9)	239 (22.5)	244 (20.5)	0.2 (.07–.4)
ICU admission	42 (32.8)	633 (59.5)	675 (56.7)	0.6 (.4–.7)
Prior systemic antibacterials in the 14 d before culturee	98 (76.6)	885 (83.3)	983 (82.5)	0.9 (.8–1.0)
Prior systemic antifungals in the 14 d before culturee	4 (3.1)	122 (11.5)	126 (10.6)	0.3 (.1–.7)
Central venous catheter	76 (59.4)	714 (67.2)	790 (66.4)	0.9 (.8–1.0)
Previous candidemia	9 (7.0)	79 (7.4)	88 (7.4)	0.9 (.5–1.8)
Patient location/Epi class	n = 128	n = 1063	N = 1191
Location prior to current hospital admission
Private residence	85 (66.4)	765 (72.0)	850 (71.4)	0.9 (.8–1.1)
Hospital inpatient	7 (5.4)	60 (5.6)	67 (5.6)	1.0 (.5–2.1)
Long-term care facility	11 (8.6)	163 (15.3)	174 (14.6)	0.6 (.3–1.0)
Long-term acute care hospital	1 (0.8)	14 (1.3)	15 (1.3)	0.6 (.08–4.5)
Homeless	17 (13.3)	9 (0.8)	26 (2.2)	15.7 (7.1–34.5)
Incarcerated	1 (0.8)	3 (0.3)	4 (0.3)	2.8 (.3–26.4)
Other	1 (0.8)	23 (2.2)	24 (2.0)	0.4 (.5–2.7)
Unknown	1 (0.8)	6 (0.5)	7 (0.6)	1.4 (.2–11.4)
Patient group/Epi class
Hospital onset	52 (40.6)	566 (53.3)	618 (51.9)	0.8 (.6–1.0)
Hospital acquired, community onset	47 (36.7)	419 (39.4)	466 (39.1)	0.9 (.7–1.2)
Community associated	29 (22.7)	78 (7.3)	107 (9.0)	3.1 (2.1–4.5)
Disposition/mortality	n = 128	n = 1063	N = 1191
Length of stay, d, median (IQR)	11.5 (5–28)	17 (8–34)	16 (7–34)	…	.032c
Time from admission to culture, d, median (IQR)	0 (0–11)	4 (0–15)	3 (0–15)	…	.006c
Overall mortality	13 (10.2)	312 (29.4)	325 (27.3)	0.34 (.2–.6)
In-hospital mortality	11 (8.6)	292 (27.5)	303 (25.4)	0.3 (.2–.6)
At 48 h (from time of culture)	6 (4.7)	89 (8.4)	95 (8.0)	0.6 (.3–1.3)
Time from culture to death, d, median (IQR)	3 (1–6)	7 (2–17)	6.5 (2–16)	…	.023c
Other characteristics	n = 128	n = 1063	N = 1191
Additional organisms	40 (31.3)	170 (16.0)	210 (17.6)	2.0 (1.5–2.6)

Data are presented as no. (%) unless otherwise indicated.

Abbreviations: CI, confidence interval; Epi, epidemiological; HIV, human immunodeficiency virus; ICU, intensive care unit; IQR, interquartile range; NA, not available; RR, relative risk; TIA, transient ischemic attack.

^aDemographics are presented at the patient level.

^bReference group: candidemia patients without history of IDU.

^cDenotes significant results; the level of significance was set at α = .05.

^dClinical characteristics are presented at the case level. A patient may have 1 or more incident cases of candidemia.

^eIncludes day of culture.

IDU cases were less likely than non-IDU cases to have diabetes (14.8% vs 35.2%; RR, 0.4 [95% CI, .3–.6]) or malignancies (7.0% vs 29.4%; RR, 0.2 [95% CI, .1–.5]) (Table 1). IDU cases were more likely than non-IDU cases to have hepatitis C (54.7% vs 6.4%; RR, 8.5 [95% CI, 6.5–11.3]), human immunodeficiency virus (HIV; 5.5% vs 2.0; RR, 2.8 [95% CI, 1.2–6.4]) and to be homeless (13.3% vs 0.8%; RR, 15.7 [95% CI, 7.1–34.5]).

IDU cases were less likely to have had abdominal surgery in the 90 days before candidemia (3.9% vs 17.5%; RR, 0.2 [95% CI, .09–.5]), receive total parenteral nutrition in the 14 days before candidemia (3.9% vs 22.5%; RR, 0.2 [95% CI, .07–.4]), and receive care in an ICU (32.8% vs 59.5%; RR, 0.6 [95% CI, .4–.7]). There were no significant differences between IDU and non-IDU cases in exposure to antibacterial agents (76.6% vs 83%; RR, 0.6 [95% CI, .8–1.0]) during the 14 days before incident culture or with presence of a CVC within 2 days of initial culture (59.4% vs 67.2%, RR, 0.9 [95% CI, .8–1.0]).

IDU cases were twice as likely as non-IDU cases to have polymicrobial blood cultures, that is, another pathogen isolated along with Candida on the day of positive Candida blood cultures (31.3% vs 16.0%; RR, 2.0 [95% CI, 1.5–2.6]). The most common accompanying organisms among IDU cases included Enterococcus faecium and Enterococcus faecalis (5 cases each), Acinetobacter baumannii (4 cases), and Staphylococcus epidermidis, Serratia marcescens, and Stenotrophomonas maltophilia (3 cases each). Candida spp distribution differed between IDU and non-IDU cases (Table 2). Candida albicans was the most common species for both IDU (45.3%) and non-IDU cases (38.1%). Candida glabrata was less common among IDU cases: 16.1%, compared with 33.6% in non-IDU cases. However, the proportion of Candida parapsilosis was similar among IDU (16.9%) and non-IDU cases (14.0%). A similar proportion of IDU (7.0%) and non-IDU cases (7.4%) had recurrent candidemia (RR, 0.9 [95% CI, .5–1.8]).

Table 2.

Candida Species Isolated From Blood Cultures of Injection Drug Use (IDU) and Non-IDU Candidemia Cases Among Emerging Infections Program Candidemia Sites, 2017

Candida spp	Injection Drug Use (n = 137)	No Injection Drug Use (n = 1082)	Total (N = 1219)
Candida albicans	62 (45.3)	412 (38.1)	474 (38.9)
Candida glabrata	22 (16.1)	364 (33.6)	386 (31.7)
Candida parapsilosis	19 (16.9)	152 (14.0)	171 (14.0)
Candida tropicalis	11 (8.0)	66 (6.1)	77 (6.3)
Candida dubliniensis	9 (6.6)	30 (2.8)	39 (3.2)
Candida krusei	1 (0.73)	20 (1.8)	21 (1.7)
Candida, other spp	6 (4.4)	12 (1.2)	18 (1.5)
Candida guilliermondii	5 (3.6)	8 (0.74)	13 (1.1)

Data are presented as no. (%). Results show all species isolated for cases, not just initial culture. Cases can involve >1 Candida spp.

The median time from admission to positive Candida culture was 0 days for IDU cases (interquartile range [IQR], 0–11) compared with 4 days (IQR, 0–15) for non-IDU cases (P = 006). IDU cases were 3 times as likely as non-IDU cases to be community associated (22.7% vs 7.3%; RR, 3.1 [95% CI, 2.1–4.5]). Length of hospitalization was significantly shorter (median, 11.5 vs 17 days; P = .032) and risk of in-hospital mortality was lower (8.6% vs 27.5%; RR, 0.3 [95% CI, .2–.6]) among IDU cases.

DISCUSSION

In 2017, 1 in 10 cases of candidemia identified in multisite US surveillance occurred in a patient with recent IDU; the proportion of IDU-associated cases was even higher (1 in 3) in patients 19–44 years of age. IDU-associated cases accounted for more than a quarter of all candidemia cases in 2 surveillance sites. Moreover, based on data from 4 sites with multiple years of data, the proportion of candidemia cases that occurred among patients with recent IDU history more than doubled from 2014 to 2017. Although IDU is a known risk factor for candidemia, the proportion of IDU cases was unexpectedly high, as candidemia is generally a healthcare-associated infection. Past month heroin use and past year heroin use disorder among those aged 18–25 years has been increasing since 2008 [9]. The proportion of past year heroin use disorder among this age group dropped slightly in 2015 and 2016 but has risen again in 2017 [9]. Additionally, the proportion of those aged 26 years and older with past month heroin use has also increased between 2002–2015 and the proportion of users in 2017 was higher than in most years between 2002 and 2015 and was similar to 2016 [9]. The dramatic increases in rates of opioid and IDU (76% increase of admissions attributed to IDU from 2004 to 2014) [10] and overdose deaths (eg, 300% increase in heroin overdose deaths between 2011 and 2015 and 9.8% increase from 2016 to 2017) [13, 14] across the United States likely play a major role in the changing epidemiology of candidemia in this decade.

A 2018 publication reported increases in Candida bloodstream infections related to injection drug use during 2009–2016 at a single hospital in Massachusetts [15]. Our analysis of population-based surveillance data from 9 sites across the United States suggests that this problem is widespread, albeit with substantial variability by site, with the proportion of candidemia cases involving IDU exceeding 10% in 4 of our surveillance sites.

IDU cases were distinct from non-IDU cases in many ways. Patients with IDU history were significantly younger and more likely to be non-Hispanic white. This is particularly notable given that overall candidemia incidence is 2–3 times higher among blacks than whites [16]. This may not be surprising given white non-Hispanic Americans have a higher rate of drug overdose deaths involving heroin and synthetic opioids other than methadone (6.1 per 100 000) than other races [14].

IDU cases also lacked many of the typical risk factors for candidemia, suggesting that the practice of injecting drugs itself and related factors such as being homeless may be contributing to the risk for candidemia. Nearly a quarter of IDU cases were categorized as community-associated, meaning they did not involve the typical healthcare exposures that might have put them at risk for candidemia. Previous descriptions of IDU-associated candidemia have implicated unsafe injection practices, licking needles before insertion, licking injection sites to “clean” the skin [7] and thereby introducing Candida from the oral flora, and use of lemon juice contaminated with C. albicans [17]. That polymicrobial blood cultures were seen more frequently in our IDU cases is another indication that nonsterile injection practices may be contributing to these infections. Even if the practice of IDU does not directly cause candidemia in all patients who inject, it is possible that hospitalizations for IDU-related events like overdose, trauma, or other infections that required intensive care, medical interventions, and long-term indwelling central lines, all classic candidemia risk factors, put these patients at risk for healthcare-associated candidemia.

The species distribution of Candida among IDU and non-IDU cases differed. IDU cases had lower proportion of infections caused by C. glabrata, likely because C. glabrata colonization and subsequent infection increases with age and is associated with longer hospitalization and more medically complex patients [18, 19]. Although C. parapsilosis, which commonly colonizes skin and could enter the bloodstream through inadequate skin preparation before injection, was more commonly isolated in IDU vs non-IDU cases in a recent single-site analysis [15], it was not significantly more common among IDU in our surveillance. Candida parapsilosis infections are also associated with the presence of a CVC, and it is possible that we did not detect a difference in proportion of infections caused by this species in IDU and non-IDU cases because both groups had a high proportion of cases with CVCs in place in the days leading up to candidemia.

IDU cases were associated with lower in-hospital mortality and shorter hospital stays, perhaps because patients with IDU history tended to be younger and had fewer comorbidities. Shorter hospitalizations may also be due to the high proportion of patients with IDU history leaving against medical advice [20]. Documented mortality in this group may have been higher if we could have accounted for postdischarge deaths that may have resulted from incompletely treated infections or overdose [21].

Similar to trends seen in our analysis, recent data from active, population-based surveillance for invasive MRSA infections have shown that the proportion of these cases associated with intravenous drug use significantly increased from 3.5% to 9.2% between 2010 and 2016 [11]. People who inject drugs may have >1 type of acute bacterial or fungal infection over the course of several months and may encounter the healthcare system frequently. These encounters are opportunities to begin treatment for substance use disorder, which can prevent both overdoses and infections from other pathogens [22].

Although this study involved population-based surveillance in 9 geographically diverse US sites, covering approximately 5% of the US population, the results may not be generalizable to the entire country. The geographic variability in IDU-related candidemia underscores the importance of understanding the local epidemiology of drug use. Given that only 4 sites had data for multiple years, the generalizability of the secular trend data is even more limited. Nevertheless, the increasing candidemia incidence and proportion of cases with IDU history in East Tennessee, an area with one of the highest concentrations of opioid users and IDU in Tennessee [23], suggest that other areas with increasing IDU may have similar increases in candidemia. Another limitation of this analysis is that data were obtained from medical record abstraction, and recent IDU history may not be correctly recorded. However, the fact that >50% of patients with IDU history had hepatitis C, a condition closely tied to IDU, compared with only 6% in the non-IDU group, suggests that misclassification may not be substantial. HIV was also significantly more common among patients who injected drugs. It is also possible that increased awareness of the opioid crisis among medical staff has led to better documentation of IDU in more recent years, giving an appearance of increasing trend. We did not collect information on type(s) of drugs used and whether cases had endocarditis, osteomyelitis, or other known complication of candidemia.

Although the frequency with which a clinician diagnoses a case of IDU-associated candidemia may vary based on local epidemiology and IDU practices, clinicians nationwide should be aware that IDU-associated candidemia is likely increasing. Candidemia in a patient without traditional risk factors, especially in those aged 19–44 years, should be a signal to investigate the possibility that the patient is injecting drugs and to consider polymicrobial infections. Effective interventions for preventing candidemia in this population might include treatment of opioid use disorder including medication-assisted treatment, community-based interventions like syringe-exchange programs, education on safe injection practices, and early signs of serious infections that should prompt them to seek care. In addition, continued surveillance for changing risk factors for candidemia and trends in IDU-associated candidemia is needed.

Notes

Acknowledgments. The authors acknowledge the support of Sasha Harb and Taylor Chambers of the Georgia Emerging Infections Program; Caroline Graber, RN, Sherry Hillis, MT, MPH, and Brenda Barnes, RN, of the Tennessee Emerging Infections Program; and the participating Emerging Infections Program Laboratories.

Disclaimer. The findings and conclusions in this report are those of the author(s) and do not necessarily represent the official position of the Centers for Disease Control and Prevention (CDC).

Financial support. This work was supported by the CDC (grant number CDC-RFA-CK17-1701).

Potential conflicts of interest. L. H. reports personal fees from Pfizer, Merck, Sanofi, and GlaxoSmithKline, outside the submitted work. W. S. reports personal fees from Pfizer, Merck, and Roche Diagnostics, outside the submitted work. All other authors report no potential conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.