In Part 1 I made reference to a press release from the American Academy of Ophthalmology announcing that children’s headaches rarely indicate a need for eyeglasses. The release opens by touting a “study” presented several days ago at its annual meeting purporting to provide clear evidence that vision or eye problems are rarely the cause of recurring headaches in children, even if the headaches usually strike while the child is doing schoolwork or other visual tasks.
A statement of that nature has wide-ranging implications, intended to reassure parents that headaches are unlikely to have a visual cause (a parent blogger has promoted this interpretation) and to steer the public toward having the pediatrician decide through screening if an eye examination is necessary rather than to consult with an eyecare practitioner (CNN Health has promoted this interpretation).
The burden of proof for such claims is on the investigators, because it runs counter to gold standard published research. Acknowledged by all parties to be one of the best studies ever published jointly by Optometry and Ophthalmology regarding children’s visual symptoms, the CITT noted that 32% of all children with convergence insufficiency (CI) reported headaches occurring fairly often or always while reading or doing close work. Given that the prevalence of CI occurs to a greater degree in the population studied than strabismus, amblyopia, and high ametropia combined, and that no provision is made in vision screenings by pediatricians or school nurses to detect the condition, the press release by the AAO stands to confuse rather than clarify the role of vision conditions regarding headaches in children. The study on which it is based should therefore provide an airtight case for its conclusions. It does nothing of the sort.
The press release is based on a retrospective study of records by a team from the Albany Medical Center led by an ophthalmology resident, Zachary Roth. Before getting to the data presented, the so-called study is already suspect for two major reasons. One is that the study is a retrospective review of records rather than a prospective design. The second is that the data is based on a poster rather than a published article that has undergone a review process in a peer-reviewed journal. Let’s take a closer look at the actual poster on which the data is based, which you can download here.
1. Purpose – the authors of the poster violate the basic rules of research by stating a bias at the outset, and to compound matters a bias based on anecdotal evidence without any supporting documentation. Here is what they write:
Headache is a common complaint in children. Children commonly present to a pediatric ophthalmologist for evaluation of headache, as parents and primary care physicians often suspect an underlying ophthalmologic problem [1,2]. In particular, we have observed that many such patients have presented to our office with headaches suspected to be caused by refractive errors. We have rarely identified any significant ophthalmologic disorder, including refractive errors. While prior studies have investigated the correlation between refractive errors and headache [3, 4], no study has compared headache resolution in children receiving, vs. those not receiving, refractive correction. Our goal was to investigate this comparison among our patients.
Here is the key bias, and it bears repeating: “We have rarely identified any significant ophthalmologic disorder, including refractive errors.” That statement is contradicted by the references cited. Even the ophthalmologic source cited as reference #3, the 2008 JAAPOS study concludes: “Compound and mixed types of astigmatism, anisometropia, and miscorrection of refractive error were found more often in patients with headache than in control subjects.”
2. Methods – the authors make vague reference to consensus guidelines on how the decision was made to prescribe glasses. There is no source or reference given for these consensus guidelines, nor any way of knowing how these guidelines relate to factors known to influence headache, such as accommodation, convergence, and accommodative-convergence interaction. Here is what the authors write:
Every child underwent standard ophthalmologic assessment, including visual acuity, pupils, external or slit-lamp examination, motility, dilated fundus examination, and cycloplegic retinoscopy. On follow-up visits, older children underwent manifest refraction when appropriate. As is our practice, spectacles were prescribed when appropriate, with reference to consensus guidelines.
3. Results – The authors report that 13.9% of their population experienced headaches associated with visual tasks. This means that 86.1% of their population had a headache type for which visual input was not a factor, and therefore a severely skewed population in which one would not expect to find that glasses had more than a chance influence.
The critical research question to answer here would be to take the 13.9% of the children who had headaches associated with visual tasks, and analyze what percentage had headaches that either resolved spontaneously, or were influenced by a change in prescription. Since this represented only 22 children the data set would have been too small to draw any conclusions, but at least the authors would have had good pilot data for a real study. They chose to ignore this, and one wonders why.
4. Conclusions– the authors conclude: We hypothesize that headaches in children are rarely cause or significantly influenced by refractive errors or other ophthalmic disorders. Our data support this assertion.
This statement violates one of the basic tenets of good research, namely stating a null hypothesis and providing data that rejects the null hypothesis. The authors began with an anecdotal, clinical impression that children presenting to pediatric ophthalmology practice with a chief complaint of headache are wasting the practitioner’s and parents’ time because the headache is rarely if ever related to eye or vision issues including the need for glasses or a change in glasses. They presented no epidemiological evidence or support in the literature for this presumption.
The authors did not conduct a prospective study, and this retrospective study was a review of records that is repeatedly criticized in medical literature as subject to selection bias. This indeed occurred, as the children analyzed in this chart review represented a population for whom we would not have expected glasses to have made any significant difference regarding headaches. It is not surprising that the authors suggest that for the population they studied, migraines or sinusitis might be a root cause of headaches.
The press release based on this skewed population claims to reassure the public that all headaches experienced by children are likely due to non-visual causes or would have resolved spontaneously. The implication is that a vision examination is a blind alley, and that parents and pediatricians should look elsewhere because glasses or a change in prescription are unlikely to have any bearing on headaches.
Contrary to this, credible evidence exists that when children present to an eyecare practitioner with headaches associated with near vision tasks, vision problems such as convergence insufficiency can be a significant causative factor, and that these issues do not resolve spontaneously. Of course, it is understandable that if the children represented by the records reviewed in this poster clogged up the surgeon’s schedule with relatively low yield appointments, that would present a major headache to the doctor and staff.
In summary, on a scale of scientific level of evidence required to draw conclusions from a study, the evidence provided in this poster and the format in which it is presented ranks very low. On a marketing scale, however, it ranks very high in its thinly veiled agenda to support screenings by pediatricians in place of comprehensive eye and vision examinations by eyecare practitioners. It will likely confuse the public more than adding clarity to important questions regarding children’s health issues.