Topical ocular chloramphenicol eye drops are relatively inexpensive and have broad-spectrum coverage of most gram-positive, gram-negative, and anaerobic bacteria. Worldwide, it is estimated that more than 50% of the world use topical chloramphenicol as a first-line agent in the treatment of conjunctivitis.

The topical eye drop form of chloramphenicol became available in 1948 and the first fatality from eye drops was reported in 1955. In 1982, Fraunfelder reported another fatality associated with this medication. A subsequent research letter in 1993 reviewed all known case reports of this adverse drug reaction, with 23 blood dyscrasias leading to 12 deaths. In 2007, chloramphenicol eye drops’ relation to aplastic anemia and blood dyscrasias was classified as “probable” according to World Health Organization (WHO) criteria, based on the known published case reports and the spontaneous reports submitted to the National Registry of Drug-Induced Ocular Side Effects (Casey Eye Institute, Portland, Oregon, USA; www.eyedrugregistry.com ).

Within 2 years of the 1982 case report of death associated with topical ocular chloramphenicol, sales in the United States declined by 90% and the Physicians Desk Reference placed a black box warning that read: “…ocular chloramphenicol should not be used unless there is no alternative.” However, most countries continued to use the eye drop form of chloramphenicol, and in 2002 England made it the first-line drug of choice for conjunctivitis and it was made available over the counter without the need for a prescription.

The rarity of this adverse drug reaction makes it difficult to study with any certainty. Some patients are genetically predisposed to develop blood dyscrasias when they are prescribed chloramphenicol systemically, and it stands to reason that this would hold true for topical administration as well. Published studies from around the world estimate the incidence of idiopathic blood dyscrasia at around 1 in 2 000 000 and drug-related incidence as higher, depending on the medication. What is the incidence of blood dyscrasia or aplastic anemia from topical ocular chloramphenicol? This answer is not known; however, there is an estimation based on population studies published since the 1993 case series. We reviewed these epidemiologic population-based studies and also collected all the case reports from the literature and the spontaneous reporting databases dating from 1993-2013.

From the review, there are 45 new case reports from the literature and from the spontaneous reporting databases since 1993. The databases accessed include the Food and Drug Administration (FDA) spontaneous reporting system, the WHO spontaneous reporting database, and the National Registry at the Casey Eye Institute. Fourteen case reports were from the spontaneous reporting systems and 31 case reports from the literature. There were 7 deaths reported out of the 14 spontaneous case reports, and no subjects were taking medications associated with causing aplastic anemia. The information from the literature was incomplete and was often in foreign languages. Data on age, duration, other medications, dechallenge, rechallenge, sex, and outcomes were consistently absent.

The epidemiologic studies include publications from around the world, including Canada, Spain, Bahrain, Israel, Germany, Bulgaria, Sweden, and the United Kingdom. The most specific study in regard to the incidence of aplastic anemia was by Laporte and associates. The investigators performed a case-control surveillance of aplastic anemia in a community of 4.2 million inhabitants (Barcelona, Spain) from 1980-1995 (67.2 million person-years). A case population risk estimate was made based on sales figures of ocular chloramphenicol in the study area during the study period. Three cases and 5 controls had been exposed to ocular chloramphenicol during the study period and the adjusted odds ratio was 3.77 (95% confidence interval). Two cases had been exposed to other drugs associated with aplastic anemia. The incidence of aplastic anemia among users of ocular chloramphenicol was 0.36 cases per million weeks of treatment. The incidence among non-users was 0.04. Extrapolation from these data led to the estimate of a less than 1 per million treatment courses. Other population studies stated the number of case reports in their country and compared that number to the number of exposures. A multicenter study including patients from Sweden, the United States, Thailand, and Israel found no cases of aplastic anemia or blood dyscrasia in subjects representing 185 million person-years that had received chloramphenicol eye drops. In all of these population studies, the authors concluded that the data do not implicate chloramphenicol eye drops as causing this adverse reaction, as the incidence of idiopathic or other drug-related instances of blood dyscrasias could account for all of the known case reports.

The question posed in the title of this article—“Did we overreact?”—is rhetorical. There is no answer because it is probably impossible to prove causation between topical ocular chloramphenicol and blood dyscrasia, primarily because an extremely rare reaction such as this cannot not be practically studied. In addition, chloramphenicol aplastic anemia toxicity is unpredictable, mortality is high, and therefore marrow studies for cell toxicity are not possible except in recovery. Also, recovery may occur through evolution of a chloramphenicol-resistant clone of hematopoietic cell, in which case studies would be misleading on these subjects. Further, there is a delay between exposure to chloramphenicol eye drops and the occurrence of drug-induced aplastic anemia, with almost 80% of subjects exhibiting symptoms and signs while no longer receiving the medication. Finally, there are many other medications implicated in causing aplastic anemia, including sulfa drugs and phenothiazines, confounding the association with chloramphenicol eye drops in some instances.

From literature reviews, case reports, and book chapters on this subject, we have come to some conclusions. First, topical ocular chloramphenicol “possibly” can cause blood dyscrasias and aplastic anemia, and the latter is frequently fatal. This is different from our previous designation as “probable” using WHO criteria in prior publications. Second, it is not possible to quantify the risk, although it appears to be very rare. Quoted incidence is 1 in 1 000 000; however, these data are based on a very small number of case reports. Third, it is likely that there are genetically susceptible individuals to this idiosyncratic reaction, and it is not possible to identify who these subjects are. Fourth, if American medicine continues to tilt towards more socialized care and some medications are not available based on cost alone, drugs like chloramphenicol eye drops could be considered as a viable treatment option. Finally, it may be too late to bring chloramphenicol eye drops back into the American market; however, in the future, it would behoove clinicians and regulatory agencies to act rationally and not change prescribing habits based on single case reports or small case series.

Clinicians have the responsibility of making risk-benefit decisions multiple times a day. The medical-legal risk of prescribing topical ocular chloramphenicol exists because a patient could develop a blood dyscrasia temporally related to when the topical eye drops were prescribed. It is the authors’ opinion that this association cannot be proven beyond stating that it is “possible,” and there is no way to screen for who may and may not be susceptible to this adverse reaction. It is the duty of the prescribing physician to act in his or her patient’s best interest, and personal judgment, best available evidence, awareness of epidemiology of disease, and following the “good medical practice” rule will enable us to keep these decisions in perspective.