[Skip to Navigation]
April 18, 2019

Utility of Ophthalmologic Screening for Patients With Candida Bloodstream Infections: A Systematic Review

Author Affiliations
  • 1Vanderbilt University School of Medicine, Nashville, Tennessee
  • 2Department of Ophthalmology and Visual Sciences, Vanderbilt Eye Institute, Vanderbilt University Medical Center, Nashville, Tennessee
  • 3Department of Ophthalmology, Harkness Eye Institute, Columbia University College of Physicians and Surgeons, New York, New York
  • 4Department of Medicine, Vanderbilt University Medical Center, Nashville, Tennessee
  • 5Department of Pediatrics, Vanderbilt University Medical Center, Nashville, Tennessee
JAMA Ophthalmol. 2019;137(6):698-710. doi:10.1001/jamaophthalmol.2019.0733
Key Points

Question  How often are Candida bloodstream infections associated with sight-threatening disease (endophthalmitis), and when found, does routine ophthalmologic screening lead to favorable outcomes?

Finding  In this systematic review of data in 38 studies of 7472 patients who underwent ophthalmologic screening for candidemia or fungemia, discrepancies in the definition of endophthalmitis led to overreporting in studies with discordant definitions of endophthalmitis compared with studies with concordant definitions of endophthalmitis; none of the 7472 screened patients had histologic or microbiologic evidence of Candida or fungal organisms in ocular tissue when results were positive for endophthalmitis. When true endophthalmitis was detected, invasive treatment was associated with blindness or poor visual outcomes compared with medical management alone.

Meaning  This review suggests that indiscriminate ophthalmologic screening for Candida bloodstream infections is not warranted.


Importance  The Infectious Diseases Society of America recommends ophthalmologic examinations for everyone with positive Candida blood culture results (candidemia) to screen for endophthalmitis, a practice that remains controversial because of multiple concerns for its limited usefulness and potential for harm.

Objective  To determine guideline efficacy by reconciling discrepancies in the incidence of endophthalmitis and evaluating outcomes of studies assessing ophthalmologic screening for candidemia.

Evidence Review  PubMed literature searches, including the search terms candidemia, fungemia, chorioretinitis, and endophthalmitis, identified longitudinal studies prior to 2018 of patients who underwent ophthalmologic evaluations in the setting of positive fungal blood culture results regardless of symptoms or clinical status. Additional studies not captured by these queries were found by manually scanning references within the articles captured by the queries. Ambiguous studies of patients with concomitant bacterial or viral infections were excluded.

Findings  Thirty-eight applicable studies of 7472 patients who underwent ophthalmologic screening for candidemia or fungemia were identified. Criteria were compared with the conventional definition of endophthalmitis based on present (concordant) or absent (discordant) frank vitreous involvement. Concordant (59 of 6693 [0.9%]) and discordant (114 of 779 [14.6%]) endophthalmitis incidence rates differed by 13.8% (95% CI, 11.4%-16.4%; P < .001). Visual acuity for each case was recorded verbatim as subjective report provided by each study, when available. None of the concordant endophthalmitis cases reported direct, intraocular, microscopic evidence of Candida or other fungal organisms. Outcomes were available for 19 patients with concordant endophthalmitis; 6 died within 4 weeks of screening. The rate of substantial vision loss was associated (φ = 0.58; 95% CI, 0.01-0.86; P = .046) with additional invasive intervention (3 of 6 [50.0%]) compared with medical management alone (0 of 6).

Conclusions and Relevance  In this systematic review without meta-analysis, inconsistent definitions of endophthalmitis accounted for discrepancies of its incidence and overreporting among patients with candidemia, contributing to bias and resulting in the construction of guidelines. As few as 3 of 7472 patients had potential improvement, while routine examination overall could lead to additional interventions and harm in this population. These findings suggest that indiscriminate screening based on candidemia alone does not appear to be supported by the literature and should be reevaluated for inclusion as a recommendation from the Infectious Diseases Society of America.

Add or change institution