Portal de Periódicos da CAPES

The good (logMAR), the bad (Snellen) and the ugly (BCVA, number of letters read) of visual acuity measurement

2016; Wiley; Volume: 36; Issue: 4 Linguagem: Inglês

10.1111/opo.12310

1475-1313

David B. Elliott,

Ophthalmology and Visual Impairment Studies

The journal continues to receive many papers that have used Snellen charts as part of ophthalmic research studies, despite logMAR (log of the Minimum Angle of Resolution) charts being widely recognised as providing much more reliable and discriminative visual acuity (VA) measurements than Snellen charts1 and being long ago proposed as standard for clinical research and clinical trials of ophthalmic devices or drugs.2-4 VA measurements using a logMAR chart have been shown to be twice as repeatable as those from a Snellen chart1 and over three times more sensitive to inter-ocular differences in VA and therefore substantially more sensitive to amblyopic changes for example.5 LogMAR VA charts use the design principles suggested by Bailey and Lovie,6 including 0.1 logMAR progression of letter size7 from −0.3 to 1.0 logMAR (equivalent to Snellen 6/3 to 6/60 or 20/10 to 20/200), five letters per line, letters of similar legibility and per-letter scoring. Many logMAR charts have a 'bottom line' of −0.3 logMAR (6/3 or 20/10), whereas many Snellen charts have a bottom line of 6/5 or 20/15 and thus provide truncated data given that the average VA of a young adult is about −0.14 logMAR (Snellen 6/4.5 or 20/15).8, 9 The Bailey-Lovie6 or ETDRS10 (Early Treatment Diabetic Retinopathy Study; this study was the first to use the chart) charts are the most commonly used logMAR charts for adults.3 All logMAR charts have used the 0.1 log progression of letter size, but various other design features differ between the charts. For example, Bailey and Lovie's original chart adopted the ten 5 × 4 British Standard letters, DEFHNPRUVZ and used a working distance of 6 m;6 whereas the ETDRS charts use the ten 5 × 5 Sloan letters, CDHKNORSVZ at a standard distance of 4 m.10 The ETDRS chart has become the most widely used in clinical trials and population studies and was adopted as the gold standard for VA measurement by the International Council of Ophthalmology.2, 4 Despite its importance as the principal measure for visual function, VA measurement and reporting is variable1 and there are several areas of confusion. Many papers report measuring VA with a Snellen chart and then analyse the data after converting it to logMAR.11-14 Snellen charts have an irregular progression in size and are often truncated, so that the data may not follow a normal distribution and parametric analyses might be best avoided.15 This is not changed merely by converting the data to logMAR. The conversion of Snellen data to logMAR can also be misleading as the abstract and results sections of papers that use this process sometimes give the impression that data have been collected using logMAR charts suggesting the data are more reliable that they actually are.11 Best Corrected Visual Acuity or BCVA is common within recent literature.16 A PubMed search of 'BCVA' (on 18th May 2016) listed almost 4000 (3987) references dating back to 1995 and its use is steadily increasing year-on-year from 22 in 1995 to 110 (2000), 178 (2005), 437 (2010) and 685 in 2015. The earliest papers using this term were all investigating refractive surgery outcomes, so that the term was used to differentiate VA after surgery (the patient's new unaided VA) with the best VA possible with a subjective refraction result (measured to determine if the surgery was reducing the optimal VA, perhaps by increased aberrations or scarring etc.). The International Council of Ophthalmology used the term throughout its 2002 Visual Standards report4 and it is now used throughout the ophthalmology (and to a lesser extent optometry13, 14, 17-19) literature. Unfortunately in many cases, it would appear that by adding the superfluous term 'best-corrected' to 'visual acuity', authors of reports believe that there is no need for further description. When the refractive correction used in the measurement of BCVA is occasionally provided in the better described papers, it has been reported as the habitual or presenting correction (e.g., spectacles, contact lenses or corneal transplant for patients with keratoconus19) or an autorefraction result20 or a 'standard refraction'.13, 14 The standard refraction is typically unspecified, but it would appear that this can vary. For example, one submission to this journal included data of BCVA after subjective refraction, which was subsequently clarified as a subjective over-refraction using +0.50 DS and −0.50 DS over the patient's own spectacles21 and without the assessment of astigmatism that most optometrists would consider part of a standard subjective refraction. Particularly given that the VA with a subjective refraction can be 0.20 logMAR or two lines better than with the presenting or habitual correction in over 20% of patients,22 there is a clear need to know whether optimal or presenting/habitual VA is being presented. Decimal VA is commonly used in Europe (except the UK which uses metric Snellen) and this can be easily confused with logMAR in the region of 6/9 to 6/24 (see Table 1). It seems best avoided. The fact that logMAR VAs better than 6/6 or 20/20 are negative is counterintuitive. Two alternatives, the Visual Acuity Rating (VAR) score23 or 'the number of letters read' (see below), have been proposed to counteract this. VAR = 100 – 50 logMAR, so that 0.00 logMAR (6/6, 20/20) = 100 VAR and each letter has a score of 1. However, after a little use the logMAR system becomes relatively easy to use and understand.1 Indeed, our current 2nd year undergraduate cohort, after using logMAR scores in their first year of clinical training, much prefer it to Snellen. The great majority of studies evaluating interventions for age-related macular degeneration report VAs by the number of letters read correctly and report improvements by the number of letters gained.12, 24 This measurement scale depends totally on the chart and the working distance used and is a source of great confusion. Papers appear to use the system most clearly described by Beck and colleagues,25 which is the scores shown in Table 1, with 6/6 or 20/20 equal to 85 letters. This is the score provided by the computerized adaptation of the ETDRS method.26 To use this system with the standard ETDRS chart with five letters per line,10 you need to either measure VA at a non-standard 1 m (for patients with poor VA)25 or measure VA at the standard 4 m and add 30 letters to the number of letters read.27 This is clearly understood by researchers using the system in wet AMD studies, but the system is now permeating into standard clinics and confusion abounds. This measurement system is the standard in this area of research and has led to some strange results sections in publications. For example, in Peden et al., Snellen VAs were retrospectively collected from records and converted to letter values 'in a standard fashion for statistical analysis'.12 Graphs misleadingly include Y-axes of 'ETDRS letters' with figure 5 scores averaging around 60 letters (0.50 logMAR, Snellen 6/18, 20/60), whereas the patient has likely read about ten letters on the top four lines of a Snellen chart. Their figure 4 suggests that patient with 'blindness (≤20/200)' improved by an average 25 letters with anti-VEGF treatments, but what this actually means is guesswork given that there is no indication of how Snellen VA <20/200 was scored and subsequently converted to logMAR and 'letters read'. In addition to the submission of papers that have used Snellen VA measurements, the reporting of the detail of VA measurements in submitted papers is often poor or non-existent. Ideally, the following information should be provided with any paper reporting VA measurements1, 3, 23: