As of March 22, 2013, a total of 47 cases among 45 patients had been identified in 9 states. Twenty-three cases were confirmed through histopathologic examination or microbiological testing; of these, infection with F. incarnatum-equiseti species complex mold was confirmed by culture for 7, and a different mold, B. hawaiiensis, was confirmed by culture for 9. The rest of the cases were confirmed only by histopathology. The definition of a probable case was met by 24 cases. Two case-patients were hospitalized; none died.

CDC confirmed F. incarnatum-equiseti species complex in an isolate derived from an unopened BBG vial from the implicated lot. Fusarium isolates from case-patients and the BBG vial were indistinguishable by multilocus DNA sequencing at 4 loci. B. hawaiiensis was confirmed in 10 isolates from 9 case-patients in 3 states.

Case Summaries

Figure 1

Figure 1. Confirmed and probable cases of postprocedural fungal endophthalmitis, by state, United States, 2011–2012Infections occurred after exposure to a product from Franck’s Compounding Lab (Ocala, FL, USA), though March 2012, when the...

With epidemiologic and laboratory evidence demonstrating 2 distinct outbreaks associated with 2 Franck’s products, all 47 cases could be sorted into 1 of 2 outbreak cohorts. Overall, 21 cases of fungal endophthalmitis were associated with BBG exposure and Fusarium mold and 26 cases were associated with Franck’s triamcinolone exposure and Bipolaris mold (Figure 1).

Figure 2

Figure 2. Epidemic curve of confirmed and probable cases of postprocedural fungal endophthalmitis, by week of procedure, United States.

Dates of symptom onset were difficult to ascertain because of the subacute onset of disease; Figure 2 reports case counts by date of procedure during which exposure to BBG or triamcinolone first occurred. In this outbreak, exposure to Franck’s BBG occurred primarily during late October–late December 2011, whereas exposure to Franck’s triamcinolone occurred during December 2011–March 2012.

BBG-associated Outbreak

Data were available for all 21 BBG-exposed case-patients (Table 3). Among these case-patients, 16 (76%) were women, median age was 69 years (range 58–86), 11 (52%) had a history of diabetes mellitus 11 (52%) had a history of hypertension, and 14 (67%) had undergone previous eye surgeries or procedures. All patients in this cluster reported unusual vision loss. Vitreous debris was noted during examination of 12 (57%) case-patients; hypopyon, eye inflammation, and fibrin were observed in 9 (43%), 10 (48%), and 12 (57%) case-patients, respectively. Eye pain was reported by 16 (76%) case-patients and floaters by 4 (19%).

Antifungal therapy was received by 17 (81%) case-patients. Among all treated case-patients, 10 (59%) received combination antifungal therapy. The most common treatment was intravitreal voriconazole, received by all 17 (100%) treated case-patients (Table 3). Oral voriconazole was the next most common treatment, received by 6 (35%) treated case-patients, followed by intravitreal amphotericin B, received by 2 (12%) treated case-patients. The median time from exposure to diagnosis in this cluster was 78 days (range 60–125); 95% of patients required additional surgeries (e.g., vitrectomies) to treat their infections (Table 3). After 6 months, 3 (14%) case-patients did not have documentation of resolved infection.

Triamcinolone-associated Outbreak

Detailed data were available for 19 (73%) of 26 triamcinolone-exposed case-patients (Table 3). Among these case-patients, 8 (42%) were women, median age was 67 years (range 53–77), 15 (79%) had a history of diabetes mellitus, and 15 (79%) had a history of hypertension. Most case-patients had a history of previous eye surgery or procedures. Virtually all (18 [95%]) case-patients experienced vision loss. Vitreous debris was commonly noted on examination. Hypopyon, inflammation, and fibrin were also observed during examination of the affected eye, although with lower frequency than for patients in the BBG cluster (Table 3). Pain was reported by 4 (21%) case-patients and floaters by 6 (32%). Two patients had received bilateral injections with triamcinolone, and for each of these patients, infection subsequently developed in both eyes.

All case-patients in this cluster received antifungal therapy, most commonly with intravitreal (11 [58%]) or oral (11 [58%]) voriconazole (Table 3). Most case-patients received a combination of antifungal therapies. Intravitreal amphotericin B was received by 8 (42%) case-patients. Median time from exposure to diagnosis in this cluster was 80 days (range 35–185); 84% of case-patients required additional surgeries (Table 3). After 6 months, documentation of resolved infection was unavailable for approximately half (42%) of these case-patients; 2 required enucleation because of severe infection. Both enucleations were performed after the 6-month follow-up period and were not systematically reported. Additional data for these 2 case-patients are not available, and more enucleations or other unreported complications among case-patients are possible.

Two concurrent outbreaks of fungal endophthalmitis were associated with 2 environmental molds contaminating 2 compounded medications labeled as sterile from the same compounding pharmacy. Together, these 2 outbreaks represent the largest reported outbreak of infectious endophthalmitis and one of the largest outbreaks in the United States attributed to contamination of a compounded medication.

Infections among case-patients were characterized by poor outcomes, including vision loss, as has been described for fungal endophthalmitis (5,9,10). Prolonged time to diagnosis probably contributed to these poor outcomes; diagnostic delay might be explained by the rarity of fungal endophthalmitis and the resulting unfamiliarity of some physicians with this diagnosis, the subacute nature of early infection, or the difficulty of distinguishing true infection from the occasionally observed postprocedural sterile inflammation (19). Female sex was associated with case status among patients in the BBG cluster; however, sex differences have not been reported for prior clusters of Fusarium endophthalmitis patients (20,21). The association of diabetes mellitus with case status among patients in the triamcinolone cluster probably reflects the fact that multiple patients received these injections for complications associated with diabetes or that diabetes might have predisposed the patients to infection after exposure.

Compounded medications are a vital part of the health care delivery system, but the role of BBG in this outbreak merits focused discussion. BBG is a triphenylmethane dye in the Coomassie dye family, commonly used for protein staining in biochemical analysis (22–24). BBG became popular among ophthalmologists for ocular surgery requiring dyes (P. Dugel, pers. comm.) because of its superior safety profile (25,26) and staining capabilities, compared with FDA-approved products (27,28). Although BBG is licensed and manufactured in several European countries, it is not FDA approved for any human use in the United States and cannot be manufactured by US pharmaceutical companies for clinical indications. Instead, BBG for clinical use in the United States is provided by compounding pharmacies. Whether providers who administered BBG or the patients who received it were aware that it was not FDA approved for human use is unknown. Regardless, greater transparency and improved education among involved parties as to BBG’s regulatory status (as a medication not approved by FDA for human use), particularly at the time of purchase, might have affected the extent of its use and, subsequently, the number of infections in these outbreaks.

This investigation had certain limitations. First, although the epidemiologic association between case status and triamcinolone exposure was strong, triamcinolone from the implicated Franck’s lot was unavailable for testing (all vials had been either opened or used in their entirety when the investigation began). Second, follow-up data through 6 months after the procedure or resolution of infection, whichever occurred first, was not available for all case-patients. Third, whether a lack of documented resolution of infection at the end of the 6-month follow-up period truly represented ongoing infection for all case-patients is unclear. Detailed treatment data, including dosages, were not systematically available, and some case-patients were still receiving treatment at the end of the 6-month follow-up period, which limited our ability to correlate treatment strategy with patient outcomes. Last, objective data on visual acuity (e.g., 20/20 scale measurements) were not available for all case-patients; because some degree of temporary, minor vision loss is not unusual after any invasive ocular procedure (P. Dugel, pers. comm.), we relied on the clinical judgment of treating ophthalmologists to define atypical postprocedural vision loss.

These outbreaks of fungal infections are not the first associated with contaminated compounded products; they preceded (by only a few months) a large multistate outbreak of Exserohilum rostratum infections associated with compounded methylprednisolone acetate (29). A 2002 outbreak of meningitis and sacroiliitis attributed to compounded corticosteroid injections contaminated with Exophiala dermatitidis was also reported (30). Unlike outbreaks associated with bacteria in compounded medications, those associated with fungi present unique challenges, because the insidious, prolonged nature of certain fungal infections might obscure an association with an earlier medication exposure. In addition, unlike bacteria, fungi are often difficult to isolate by culture; molecular diagnostic techniques, such as PCR, might be more useful, but their availability to the average clinician is limited. Furthermore, when directly inoculated into sterile spaces, these fungi can cause infections that are challenging to recognize, as we observed in these outbreaks, not only because they often involve environmental agents that are exceedingly rare causes of human illness but also because once instilled into a sterile space, they can cause disease in an uncommon anatomic area or in an unusual way.

Bipolaris spp. are a common cause of chronic sinusitis among immunocompetent persons (31); however, before this outbreak, endophthalmitis with Bipolaris spp. had been described in limited case reports only (32,33). Similarly, fusariosis usually occurs in immunocompromised hosts; endophthalmitis has been reported but usually as a sequela of Fusarium keratitis (10,34) and not de novo, as in this outbreak. The rarity of the infections in these outbreaks provided an additional clue to clinicians that an atypical event had occurred, underscoring the crucial role of clinicians in recognizing unusual disease clusters and promptly reporting them to public health authorities for rapid epidemiologic investigation necessary to rule out widespread exposure.

Oversight of compounding pharmacies lies primarily with state pharmacy boards because compounding pharmacies are regulated as pharmacies despite the fact that some (e.g., Franck’s) produce large quantities of medications and distribute them across state lines. Federal oversight of compounding pharmacies has been limited and often contested. FDA had investigated Franck’s twice before these outbreaks. During 2004–2005, FDA inspected Franck’s for improper compounding practices, including compounding medications without a valid patient prescription (35), and during 2009, Franck’s veterinary compounding unit incorrectly compounded a nutritional supplement that resulted in the deaths of 21 polo horses (36). These investigations did not preclude Franck’s from continuing its operations. During 2010, FDA sought a federal injunction to compel Franck’s to cease veterinary compounding from bulk ingredients (37) but ultimately lost the court case (38). Enhanced regulatory authority toward compounding pharmacies with multiple infractions should be considered a part of efforts to improve compounded medication safety. Furthermore, greater transparency surrounding ongoing disciplinary action against a compounding pharmacy might empower clinicians to make more informed purchasing decisions. Enhanced oversight of compounding pharmacies that produce large quantities of medications, especially sterile products that are distributed nationwide, should be considered to ensure that mass-produced sterile compounded products are safe.

Clinicians should be aware that the availability of a compounded medication in the United States is not a guarantee of its quality or of FDA approval. Disclosure of a medication’s FDA approval status should be encouraged at all stages of purchase and use. This information might enable clinicians to make informed decisions about the medications they purchase for patient use and to educate patients about the status of medications to which they are exposed. Maintenance of the safety and integrity of sterile compounded drugs in the United States demands a thorough review and improvement of compounding pharmacy regulatory practices.