Critique of Creavin et al. (2015) Pediatrics paper

Guest blogger, Professor Bruce Evans FCOptom, offers his critique of the recent article Ophthalmic Abnormalities and Reading Impairment by Creavin et al

Share options

Author: Professor Bruce Evans
Date: 29 May 2015

You may have seen the stories in the media this week about dyslexia and vision, including Forget colour overlays – dyslexia is not a vision problem in New Scientist and Dyslexia not linked to eyesight, says study on BBC News. These stories followed publication of the article Ophthalmic Abnormalities and Reading Impairment by Creavin et al. Here, guest blogger Professor Bruce Evans FCOptom, offers his critique of the study.

The tests

The binocular vision tests that the authors carried out are very coarse and are mostly not the tests that are recommended to detect the visual correlates of dyslexia. Most notably, they didn’t test pattern glare or the effect of coloured filters so would not have detected the visual anomaly (Visual Stress) that I think most commonly co-occurs with dyslexia (in about a third). Indeed, the literature review section of the paper is sorely lacking in many respects. Some of the press reports say that the paper indicates that coloured filters are not helpful, which is incorrect. The paper did not investigate the effects of coloured filters nor look for Visual Stress, the condition that the coloured filters correct.

The authors also made no attempt to find out about their participants’ symptoms. Previous work has shown that subtle binocular vision and accommodative anomalies are associated with symptoms.1 Whilst these optometric factors are unlikely to be major causes of reading difficulties (RD), if they cause blurred or double vision, eyestrain, and headaches then they will be additional burdens and treatment may take away one component of the child’s difficulties.

Scoring and analysis

This is the most disappointing aspect of the Pediatrics paper. For several tests, the authors have carried out a test that produces a continuous variable and then, rather than comparing the group means (e.g., with a t-test), the authors have reduced the data to a binary (dichotomous) classification of normal/abnormal. Of course, this throws away much of the variance in the data and greatly reduces the power of the study to detect change. Even worse, where the authors have done this their cut-off for normal/abnormal seems arbitrary, with no sound rationale for the cut-off they have selected. For some variables (e.g., accommodation, convergence) the authors don’t even tell the reader where the cut-off was set. 


Considering the coarse tests that the authors used, it is noteworthy that they found abnormal ophthalmic function in 1 in 5 children with RD. For example, the RD group were 3 times more likely to “fail” the stereotest, even with the cut-off that the authors used. Despite finding some visual correlates of RD the authors go on to hypothesis that the reduced binocular functions could be the result of dyslexia although they present no data to evaluate this. This was something that has been evaluated by previous research which found that the visual correlates of dyslexia are not secondary to the poor reading.1 The Pediatrics paper goes on to say that treatments won’t help; but they did not look at treatments!

Overall, I think that previous work supports the view that the visual correlates of dyslexia are subtle and not likely to be major causes of RD, but can in some cases cause an extra burden when the child reads.2 There is nothing in the Pediatrics paper to change this view and indeed they actually found evidence to support this view, although somebody who just read their conclusions would not appreciate this.

Unfortunately, the authors’ tendency to reduce their test results to pass/fail seems to extend to their conclusions, where they only seem to consider two options: visual problems cause dyslexia or visual problems are irrelevant to dyslexia. The situation is more complex and two children with RD are likely to have a very different pattern of abilities and disabilities. In a few cases, visual factors add to the disability and then a visual treatment (e.g., glasses, eye exercises, coloured filters) will give the child better access to the extra teaching that they no doubt will also need. I fear that Creavin and colleagues will do work in this field a great disservice by taking us back to a time when there were entrenched simplistic views about whether vision is everything or nothing to do with RD.

Ophthalmic Abnormalities and Reading Impairment, Creavin et al

Professor Bruce Evans PhD FCOptom FAOO FEAOO

Professor Bruce Evans FCOptom has, for over 25 years, spent most of his working week as a practising optometrist working in a busy community optometric practice. He has also been involved in over 60 research studies investigating most areas of optometric practice. 

Bruce is currently director of research at the Institute of Optometry, visiting professor to City University and visiting professor of optometry to London South Bank University. He is a Fellow (by examination) of the College of Optometrists and holds its higher qualifications of Diploma in Contact Lens Practice and Diploma in Orthoptics. He was awarded Fellowship of the British Contact Lens Association in 2006. 

His main areas of research are children’s vision, dyslexia, orthoptics, headaches, and contact lenses. He has given more than 300 invited lectures in over 20 countries and three prestigious named lectures. Bruce has over 210 published papers and has authored five books on binocular vision and two on dyslexia and vision. He has an optometric practice in Brentwood, Essex, where he specialises in children’s vision and contact lenses. 

Return to blog listings