Systematic review of Purtscher's and Purtscher-like retinopathies

Abstract

Purpose: (1) To describe the clinical characteristics of Purtscher and Purtscher-like retinopathies, including etiologies, fundoscopic signs, results of complementary investigation, treatments, and outcomes. (2) To compare visual acuity (VA) of patients receiving corticosteroids for PuR compared with observation.

Methods: Systematic review of several databases (1980-2010): Medline, EMBASE, ISI, EBSCO, Science Direct and Google Scholar. Study selection criteria included: (A) Studies of PuR with ophthalmology assessments; (B) ≥3 of 5 diagnostic criteria of PuR; (C) Quantified VA at presentation. For quantitative assessment (purpose 2), we selected only studies that reported whether corticosteroids were administered, and with vision assessments after at least 1 month.

Results: (1) From 670 studies initially found, 40 were included (68 cases, 110 eyes): 1 prospective, 5 case series, and 34 case reports. Mean VA at presentation was 1.3 logMAR (logarithm of the minimum angle of resolution) (<20/200; range: 20/20-light perception). Purtscher flecken were underreported. Trauma and acute pancreatitis were the most frequent etiologies. There were six deaths, all with systemic associations. (2) There was no statistically significant difference between VA improvement for patients treated with corticosteroids compared with observation. Visual prognostic factors identified included etiology (pancreatitis and trauma were associated with higher probability of visual improvement) and male gender.

Conclusion: Due to limitations of case reports and series, the presented data are only useful as broad characterizations of the clinical course of PuR. Further studies, possibly including trials to assess the effectiveness of corticosteroids use, and larger prospective cohort studies, are necessary, but may not be feasible to conduct.

Figure 1

Flowchart of search strategy.

Figure 2

Risk of bias graph.

Figure 3

Mean visual acuity at presentation according to etiology. In logMAR scale, 0 is equivalent to normal VA; the higher the value of logMAR, the worse is VA.