UMassMemorial Medical Center is a 781-bed academic medical center affiliated with the University of Massachusetts Medical School and located in central Massachusetts. The study was approved by the University of Massachusetts Medical School (UMMS) Institutional Review Board (IRB). Patients >14 years of age with a positive blood culture for Candida species during January 1, 2010–January 31, 2017, were included in the study. We excluded patients <14 years of age, as they were felt to be unlikely to have a history of IVDU. The protocol approved by the IRB also excluded the study of prisoners; however, none of the patients was excluded for this reason.

We identified cases of candidemia by a query of the hospital electronic medical records (EMRs) using Theradoc (Premier Inc., Charlotte, NC, USA). We then reviewed the EMRs of the patients with candidemia for any history of IVDU; we classified cases as IVDU and non-IVDU on the basis of this chart review. We included cases in the IVDU group if there was any history of IVDU, even if remote. We reviewed admission notes, infectious diseases consult notes, psychiatry consult notes, discharge notes, and toxicology screens. We then conducted a further review of the EMRs to extract data on the epidemiology, risk factors, clinical manifestations, microbiology, laboratory findings, and outcomes. The IRB protocol forbade contacting the patients directly.

We performed Candida speciation using standard techniques, including the Vitek 2-Yeast System, Vitek-MS, and API 20C (bioMérieux, Marcy l’Étoile, France). We compared the variables in the tables between the IVDU and non-IVDU populations using the Fisher exact test for categorical variables and the Wilcoxon rank sum test for continuous variables. These tests are nonparametric and avoid the need for large sample sizes. We defined statistical significance as a p value <0.05.

Our data establish that, in the setting of the ongoing opioid epidemic, candidemia appears to have emerged as a serious clinical problem in the IVDU population in central Massachusetts. Major differences appear, however, between the cases reported in the literature that were associated with injection of brown heroin and the ones in this case series. Cutaneous, ocular, or osteoarticular involvement were commonly observed in the brown heroin users with candidiasis (8,9) but were rarely seen in the IVDU case-patients we report. In contrast, although all the cases in our study featured candidemia by definition, blood cultures were rarely positive for candidemia in the brown heroin cases.

As discussed earlier, in the brown heroin cases, the infections were linked to C. albicans contamination of lemon juice used to solubilize the heroin; blood cultures, when positive, grew only C. albicans (8,9). In our case series, C. albicans was the third most common species, behind C. glabrata and C. parapsilosis. The finding of different species of Candida argues against a common source of infection. Rather, endogenous sources of Candida seem likely, because this fungus can colonize the skin and mouth; injection drug users may lick their needles before insertion into the skin. However, our IRB protocol did not permit us to interview subjects regarding injection habits or to culture their drugs and injection paraphernalia. Thus, the source of the Candida species isolated from the patients is speculative, and future studies are needed to address this question.

C. glabrata and C. parapsilosis are typically thought of as nosocomial pathogens, with spread of the pathogen facilitated by contaminated inanimate surfaces and person-to-person contact (15,16). However, both species can also be gut colonizers and commensals of human skin (16). C. glabrata is often resistant to fluconazole; a risk factor for infection with this species is prior receipt of fluconazole (15). However, none of our patients was known to have received fluconazole before contracting candidemia.

In our study, we found that the candidemia patients in the IVDU group were younger and more likely to be co-infected with hepatitis C than were patients in the non-IVDU group. This finding likely reflects the relatively young age of the IVDU population and the high prevalence of hepatitis C infection for this group (17). Even though more than half of the patients in the IVDU group were co-infected with hepatitis C, no patient had HIV infection. This finding is consistent with a 2012 report documenting declining rates of HIV infection and increasing rates of hepatitis C among injection drug users in Massachusetts (18).

In a recent multicenter study of >2,000 patients with candidemia, C. albicans was responsible for 46% of the cases, followed by C. glabrata (26%), C. parapsilosis (16%), C. tropicalis (8%), and C. krusei (3%) (1). In our study, the species distribution in the non-IVDU population with candidemia was very similar, with slightly more than half the cases attributable to C. albicans (Table 4). In contrast, only 7 (29%) of 24 candidemia cases in the IVDU population were caused by C. albicans, and 1 of those cases was a co-infection with C. glabrata. Perhaps reflecting these patients’ younger age and fewer concurrent conditions, the mortality rate was lower for the IVDU population compared with the non-IVDU population, although the difference was not statistically significant. Nevertheless, 5 (21%) of 24 patients in the IVDU group died, emphasizing that candidemia in the IVDU population is a serious disease associated with a serious risk of death.

Our retrospective study does not shed light on the optimal therapy for IVDU-associated candidemia. However, it seems reasonable to begin to treat patients with an echinocandin, according to the Infectious Diseases Society of America clinical practice guideline for the treatment of candidemia (19), pending identification and susceptibility testing of the isolate. One third of our IVDU-associated candidemia cases were caused by C. glabrata which, as noted earlier, is often resistant to fluconazole (15). In addition, the superiority of an echinocandin over fluconazole was demonstrated in a multicenter trial of patients with candidemia (20). However, a disadvantage of the echinocandins is their need to be given intravenously, a particular problem in the patient population with IVDU, because they may use intravenous lines for drug injection. In clinically stable patients with susceptible isolates and without endocarditis, transition to fluconazole can be considered when the patient is ready for discharge.

Several limitations should be taken into account when interpreting the data we present. The study was retrospective, and because of our IRB protocol, we were not permitted to contact the patients directly to verify their histories or obtain additional information. Moreover, it is often difficult to obtain an accurate history of IVDU from patients. Thus, because it was not known whether the injection drug use was current or not, we cannot be certain that the IVDU contributed to the risk of developing candidemia. Indeed, it is possible that some of the patients were misclassified and there may be inaccuracies in variables such as when the patients last injected drugs and what drugs were injected. For this reason, when designing the study, we prospectively decided to include all patients with a history of IVDU, however remote, in the IVDU group. Second, although it is very likely that the trend toward increasing numbers of IVDU-associated candidemia is real, it is possible that the way IVDU is recorded has changed over the years because healthcare providers are now more attuned to the opioid epidemic and thus record the history of IVDU more often than they did in the past. Third, although the number of cases was relatively large, the study was not powered to detect differences in uncommon events between the IVDU and non-IVDU groups. In addition, multivariate analyses were not performed, so some statistically significant associations may be confounded. Fourth, there may have been a bias toward ordering certain tests, such as hepatitis C serology and echocardiography, in the IVDU-associated candidemia group. Finally, our study was a single-center study from a tertiary care institution. It remains to be seen whether candidemia in IVDU is mostly a local phenomenon peculiar to central Massachusetts or it is emerging at other medical centers in other geographic locations.

These limitations notwithstanding, our findings establish candidemia as an underappreciated outcome of the opioid epidemic gripping the United States. Moreover, we observed a marked increase in the number of cases during the last year of the study. Although this finding may reflect random variations, it is of concern that the problem may be worsening. In Worcester County, where UMassMemorial Hospital is located, the number of opioid-related overdose deaths increased every year during the 7-year study period, from 80 in 2010 to 251 in 2016 (7). Although many of the press reports surrounding illicit opioid use have centered on deaths from overdose, infections remain a serious cause of illness and death in persons with IVDU. Invasive candidiasis must be considered in the differential diagnosis of patients with infectious sequelae of IVDU.