Ocular Findings in Children with Headache

ABSTRACT Purpose To determine the prevalence of ophthalmological findings suggesting an ocular cause for headache or occult neurological disease, among children with headache. Methods Retrospective cross-sectional study on children with headache at a tertiary outpatient ophthalmology clinic. All children underwent sensorimotor, anterior segment, and dilated fundoscopic examinations, with or without cycloplegic refraction. Prevalence of one or more new findings of ocular or occult neurological cause of headache, including glaucoma, uveitis, optic nerve elevation, or possible asthenopia from strabismus or refractive issues. Headache characteristics and associated symptoms were evaluated as risk factors for ocular findings. Results Among 1,878 children with headache (mean age 10 yrs, range 2–18), 492 (26.1%, 95% CI 24.3–28.2%) children had one or more new ocular findings that could cause headache or indicate intracranial disease: refractive issues (342, 18.2%), strabismus (83, 4.4%), optic nerve elevation (51, 2.7%; 26 with papilledema, 25 with pseudopapilledema), uveitis (6, 0.3%), and glaucoma (2, 0.1%). Shorter headache duration was associated with ocular findings (p = .047), but headache frequency, photophobia, nausea/vomiting, and visual changes were not. In univariable analysis, visual changes (p ≤ .001), nausea/vomiting (p ≤ .002), and morning headache (p = .02) were associated with optic nerve elevation. Conclusion An ophthalmologic examination including cycloplegic refraction is indicated in children with headache, as one-quarter have a treatable ocular condition, which may be related to the headache, or sign of intracranial pathology. While nausea, visual changes, or morning headache should raise concern, coincident visual, ocular, or systemic symptoms are not reliable predictors of discovering ocular pathology in a child with headache.

Headache is the most common manifestation of pain in childhood. 1 The prevalence of headache in children and adolescents is estimated to be roughly 60%. 2 In addition to being a possible symptom of a life-threatening intracranial process, headaches can impact a child's quality of life, decrease functional status, and impact school attendance and performance. 3 Children with headache are commonly evaluated by ophthalmologists. In fact, ophthalmologists have been found to be the second or third most consulted specialist in the workup of an acute onset of headache. 4,5 While some headaches may represent primary headache disorders, the International Classification of Headache Disorders 3 rd edition identifies a subcategory of headache related to disorders of the eye, such as those due to angle-closure glaucoma, refractive-error-related asthenopia, or ocular inflammatory disorders. 6 Furthermore, headache can be a sign of an occult neurologic process, and an ophthalmologic exam can reveal helpful diagnostic signs such as optic nerve elevation.
Despite the common presentation to ophthalmologists, it is not clear what the prevalence of ocular findings are in children with headache, nor is it clear what the utility of ophthalmological consultation is for such children. A study of inpatient pediatric ophthalmology consults found that 1 in 7 patients who had headache had ocular disease. 7 Studies of outpatient ophthalmologic headache workups have included both adults and children and did not stratify results by age, limiting their generalizability to the pediatric setting. [8][9][10] To our knowledge, no studies have evaluated the prevalence of ocular findings among just children with headache in an outpatient setting, where many pediatric headaches are evaluated and in which there is less likely to be a selection bias for more severe headache, as there may be among inpatient studies.
We sought to determine the prevalence of ophthalmological findings that might suggest an ocular cause for headache or an occult neurological disease, among children with headache. We also examined headache characteristics and associated symptoms as risk factors for the identification of ocular findings on examination.

Methods
We conducted a single-center, retrospective crosssectional study at the Children's Hospital of Philadelphia, Philadelphia, PA (CHOP). The study was approved by, and waiver of informed consent was granted by, the CHOP Institutional Review Board. Eligible subjects were children age 18 years or younger, with complaint of headache and were evaluated at the outpatient ophthalmology clinics between May 20, 2009, andSeptember 20, 2013. CHOP eye clinics provide primary pediatric ophthalmology care in urban, suburban, and rural settings. We included both new and established clinical patients but considered only visits at which headache was reported as a symptom by the child or caregivers. Children were excluded if they had known papilledema prior to complaint of headache; head trauma, eye trauma, or eye surgery within one month prior to the visit, or inadequate information in the record.
Data were abstracted from the medical record from the first clinical encounter that had a documented complaint of headache and included: demographics, headache characteristics (frequency, duration, time of day, and relation to reading or school work); presence of visual changes (e.g. visual aura or transient visual obscurations); presence of photophobia or phonophobia; past ocular and medical history; results from diagnostic testing (i.e. lumbar puncture (LP) and neuroimaging); and the complete ophthalmologic exam, such as visual acuity, intraocular pressure, pupillary exam, ocular motility and alignment, anterior segment examination, dilated fundus examination, and cycloplegic refraction with retinoscopy and/or phoropter following administration of 1% cyclopentolate and 1% tropicamide or 1% tropicamide and 2.5% phenylephrine. All ophthalmologic examinations were completed by a pediatric ophthalmologist or pediatric neuro-ophthalmologist.

Statistical analysis
The primary outcome was the prevalence of children with one or more new ophthalmologic findings that might indicate a possible ocular or neurologic cause of headache. The presence of a finding was considered a positive outcome regardless of whether subsequent work-up revealed a true neurologic disease or nonneurologic cause of the finding (e.g., nerve elevation determined to be pseudopapilledema after a diagnostic work-up was completed), because the study question was the prevalence of potentially significant ocular findings. Secondary outcomes were the prevalence of each individual finding. The primary and secondary outcomes were calculated for the entire study cohort of children with headache. Subgroup analysis also was completed by dividing the cohort into three subcategories based on past medical history: no known history of past ocular or intracranial pathology, known history of ocular pathology, and known history of intracranial pathology. For children with known history of ocular findings, only a new finding would count as a positive outcome; for example, a child with high refractive error presenting with a new complaint of headache, found to have an unchanged refractive error, and wearing the correct glasses, would not be considered to have a positive outcome with regard to the study.
Significant refractive error was determined to be a new finding by identifying significant discrepancy between the child's current glasses prescription or lack of glasses and the results of the cycloplegic refraction performed at the visit. Both spherical equivalent and astigmatic refractive error discrepancies were assessed for both eyes of each child, and the greater absolute value of discrepancy from either eye was used for classifying each child as having or not having a significant refractive issue using the schema described below, which is based on general consensus judgment of CHOP ophthalmologists. Significant discrepancies in anisometropic refractive correction were identified as well. If the child was wearing myopic glasses, then they were determined to have a significant refractive error if they were under-corrected (under-minused) by ≥3 diopters (D) or over-corrected (over-minused) by ≥0.75 D. If the child was not wearing glasses currently, they were determined to have a significant refractive error if they had ≥3 D of uncorrected hyperopic or ≥1 D of uncorrected myopic refractive error. If the child was wearing hyperopic glasses, then they were noted to have a significant refractive error if they were under-corrected (under-plused) by ≥3 D or over-corrected (over-plused) by ≥1 D. A significant change in astigmatism was noted if the absolute value of the change in plus cylindrical refractive error was ≥1 D. Anisometropic differences were calculated by comparing the differences in spherical equivalent refractive error between the eyes of the current glasses to that of the cycloplegic refraction. A significant discrepancy in anisometropia was noted if the absolute value of the change in anisometropia was ≥1 D.
Univariable analysis for risk factor was conducted to identify historical findings that might be associated with an increased risk of having any ophthalmologic abnormality (primary outcome) and with each individual type of ophthalmologic abnormality (secondary outcomes). We made the assumption that if a specific characteristic headache was not commented on in the medical record history, then it was absent. Descriptive statistics were calculated with means (SD) and median (range) for continuous variables. T-tests were used to compare means, and chi-square tests were used to compare proportions. In addition to chi-square tests, when counts in categorical variables were less than or equal to 5, Fisher's exact test was used. A two-sided p-value less than 0.05 was considered statististically significant. Adjustment for multiple comparisons was not made for the risk factor analyses, because these were considered secondary analyses. All analyses were performed in SAS version 9.4 (SAS Institute Inc., Cary, NC).

Results
In total, 2,463 children with headache were evaluated in the outpatient clinics during the study period. 584 children were excluded due to known papilledema (n = 29), recent trauma (n = 56), post-operative pain (n = 3), or insufficient medical records data (n = 494). Therefore, 1,878 children were included in the study ( Table 1). The mean age was 10.0 years (SD 3.7) with a range of 2 to 18 years, and 1,061 (56.5%) children were male. 92.0% had headaches as their chief complaint, and 8.0% were there for another reason, such failed vision screen or screening for diabetic retinopathy. 1,370 (72.9%) children had no previously known history of ocular or intracranial pathology, 473 (25.2%) children had a history of ocular pathology and presented with a new complaint of headache, and 35 (1.9%) children had a history of intracranial pathology and presented with a new complaint of headache. 823 (43.8%) children were referred for evaluation by their primary physician, 280 (14.9%) by a neurologist, 75 (4.0%) by the Emergency Department, 47 (2.5%) by an optometrist, and 81 (4.3%) by other specialists (endocrinologist, rheumatologist, plastic surgeon, otorhinolaryngologist, or oncologist). Seventy-seven (4.1%) children were self-referred by their parents; and 500 (42.6%) children were established ophthalmology patients presenting for a scheduled follow-up visit.
Overall, 492 (26.1%, 95% CI 24.3 to 28.2%) children had one or more new findings that may indicate an ocular or neurologic cause of headache ( Table 2). The average age of a child with any finding was 10.3 years, and 56.5% were male ( Table 3). The most frequent finding was a refractive issue, identified in 342 (18.2%) children. However, not all children in the cohort had a full cycloplegic refraction recorded in the medical record; among the 1,632 (86.9%) children who had a full cycloplegic refraction, 342 (21.0%) of these children had a refractive issue. These issues included over or undercorrected spherical myopic, spherical hyperopic, astigmatic, or anisometropic refractive error of a high enough magnitude to possibly be causing asthenopia, and these errors were found both in children who wearing glasses and had a change in their refractive error and in children who were not yet wearing glasses (Online Supplement Table 1). In addition, two children with anisometropic astigmatism had glasses that were incorrectly made; in one case, the magnitudes of astigmatism were switched between the eyes, and in the other case, the axes of astigmatism were switched between the eyes.
Risk factor analyses are presented for associations with specific ocular outcomes for headache characteristics among medical history subgroups (Table 4), and headache characteristic associations with specific ocular outcomes (Table 5). In the risk factors analysis of visual changes, photophobia or phonophobia, morning headache, end of day headache or relation of headache to reading or school work, headache frequency (daily to weekly, less than weekly, or once), and headache duration (1 month or less, 1 month to 1 year, or greater than 1 year); only shorter headache duration had an association with the presence of one or more new ocular findings (p = .047) ( Table 4). In the medical history subgroup analysis, this association between shorter headache duration and one or more new ocular findings was again the only significant finding and only present among children with no known ocular or intracranial pathology (p = .03, Table 4). Associations of risk factors with individual specific ocular findings are presented in Table 5. Photophobia, nausea/vomiting, and visual changes were not associated with ocular findings. Visual changes (p ≤ .001), nausea/vomiting (p ≤ .002), and morning headache (p = .02) were associated with optic nerve elevation. Comparing children with papilledema to those with pseudopapilledema, the former were more likely to have a history of nausea/vomiting (54% v. 24%, p = .03), but they did not differ with regard to visual changes, photophobia or phonophobia, or morning headache.

Discussion
We found that nearly a quarter of new or established children at a tertiary ophthalmology practice with headache had one or more new ocular findings on examination that may suggest an ophthalmologic source of headache or an intracranial process. This finding suggests that an examination that includes a cycloplegic refraction and a dilated fundus examination may prove helpful in the diagnostic evaluation of children with headache. There are very few studies to which to compare our findings. Fasih et al. studied 379 adults and  (15)    None None None None children (mean age 35 years, range 6-75) presenting to an outpatient ophthalmology clinic in Pakistan and reported 52% had an ophthalmological finding. 8 Asthenopia was the most frequent possible cause of headache, among which the causes including refractive errors (16%), presbyopia (15%) (which by definition was not present in children), and computer use associated asthenopia (5%); they also identified corneal ulcers (6%), angle closure glaucoma (4%), and papilledema (0.5%). 8 However, these findings were not stratified by age, and it is unclear if cycloplegic refractions were performed in the children. Another study of over 900 patients with headache (mean age 30 ± 13.5 years) in the Democratic Republic of the Congo in the 1980ʹs reported 64% had an ocular finding, of which 44% had refractive disorders, 17% posterior segment lesions, 12% conjunctiva abnormalities, 12% anterior segment abnormalities, 11% presbyopia, and 4% ocular motor palsies, heterophoria, convergence insufficiency, and other ocular abnormalities. 9 In this study, subgroup analysis for children was not available, and it is not clear if all the ocular findings, such as refractive errors, were potentially related to headache. O'Donnell found that of 18 of 63 patients (ages unreported) who presented to an eye Emergency Department with headache had an ocular finding. 10 Pure pediatric data are very limited. One small study of inpatient pediatric ophthalmology consultations found that one of only seven studied children with headache had an ocular finding. 7 One of the biggest concerns when a child presents with new onset headache is an occult intracranial process such as a brain tumor. Brain tumors are the most common solid tumor in childhood, and the most common symptom at diagnosis is headache, occurring in 33-41% of children. 11,12 We found 51 children (2.7%) presenting with headache had optic nerve elevation. Visual changes, nausea/vomiting, and morning headache were risk factors associated with optic nerve elevation. Of the children who were found to have true optic disc swelling, 15 children had pseudotumor cerebri, 1 child had a brain tumor (pilocystic astrocytoma), and seven others had other causes of increased intracranial pressure (cerebritis, Chiari 1 malformation, central nervous system vasculitis, venous sinus thrombosis, subdural hematoma, and 2 children with VP shunt malfunction). A prior study of children with pseudotumor cerebri found that nearly 90% of children with definite pseudotumor cerebri had headache, which could be either episodic or constant/ daily, and sometimes pain localizing to the head/neck/ shoulders. 13 Kovarik et al. found that 17 of 26 children being evaluated for papilledema versus pseudopapilledema complained of headache. 14 The most common finding in our cohort was a refractive issue, identified in 18.2% of children with headache in a tertiary pediatric ophthalmology clinic. Prior studies have reported associations between refractive errors and headaches. [15][16][17][18] Asthenopia, which may be characterized by chronic headache and eye fatigue, can be caused by uncorrected refractive errors or impairment of convergence. 19 In our study, among children who had a full cycloplegic refraction, 6.9% had a significant change in spherical equivalent refractive error, 12.9% had an astigmatic change ≥1D, and 8.0% had an anisometropic change ≥1 D. Two children had either their astigmatic axes or amounts switched between the eyes. Thus, even in children with a recent refraction, if there is a new onset of headache, it is important to consider refractive issues as a potential cause. Further, children may be more likely than adults to describe symptoms of asthenopia as a headache, as they may not have the words to report eye strain, and as with other diseases in children, it is important to think more broadly as to what a child means when they report head pain. Convergence insufficiency can cause difficulty with prolonged reading and near work, which can contribute to headaches particularly in school-aged children. We found that children complaining of headache with reading or at the end of the day were more likely to have strabismus on exam (p = .02), 4.4% of children were diagnosed with new or worsening strabismus when they presented with headache, and 25.3% of those with new strabismus had convergence insufficiency.
Less common but potentially more serious ocular disorders diagnosed in our cohort of children included uveitis, glaucoma, and extra ocular inflammatory conditions such as episcleritis, orbital inflammatory syndrome, and orbital cellulitis. These conditions are welldescribed causes of ocular pain and headache in adults, and are commonly listed as part of the differential diagnosis for periorbital headache. [19][20][21][22] While based on our study the prevalence of such diseases appears to be low in children, an ophthalmologic exam is still critical, as these diseases are intervenable and treatable entities, some of which may be vision threatening.
Strengths of our study included a large cohort size and eye examinations completed by a pediatric ophthalmologist or neuro-ophthalmologist. However, there are limitations to consider. This study was performed at a single institution. CHOP is a tertiary referral center, resulting in a population that may be enriched with regard to medical complex patients; however, the majority of patients are referred from primary care providers, very similar to the population that might present to community pediatric ophthalmologists. In this retrospective study, we did not establish definitive causes of headache for each child through assessing for resolution of headache after ophthalmological intervention or formal neurologic assessment for primary headache disorders, such as migraine. Nevertheless, the ophthalmologic findings were clinically significant, resulting in intervention such as prescribing glasses, treatment of convergence insufficiency and other strabismus as well as uveitis or other visual threatening disease, and workup for possible papilledema. Another consequence of retrospective data collection was that we did not perform a comprehensive risk factor inventory or formal, in-depth history for headache characteristics or associated symptoms. We assumed the absence of a headache characteristic if it was not specified in the medical record history; this assumption might have led to an ascertainment bias. While most children underwent a cycloplegic refraction, about 13% did not. For our primary outcome, these children were assumed to not have significant refractive errors, so the observed prevalence of ophthalmological findings may be an underestimate. However, such bias would not change our study conclusions, and we did report a subgroup analysis of only children with refractions. We also chose thresholds to define a "significant refractive error" based on our clinical judgment as to visually significant refractive errors and did not clearly establish that these levels of uncorrected refractive error might cause asthenopia. Other clinicians may disagree with these levels, and associating specific levels of refractive error with asthenopia requires further study. Lastly, the study performed many comparisons and correction for multiple comparisons was not applied, thus some significant associations could be due to chance. Future studies are needed to validate these significant findings. Generally, a more systematic, in-depth evaluation of headache characteristics would help to confirm or refute our observations with regard to these factors.
In conclusion, over a quarter of children with headache had one or more new ocular findings that might suggest a potential source of headache or an intracranial process. Over half of these were refractive issues. Therefore, a full ophthalmologic examination, including cycloplegic refraction, is indicated in the diagnostic workup of children with headache. While complaints of nausea/vomiting, visual changes, or morning headache should raise concern, generally the presence or absence of coincident visual, ocular, or systemic symptoms are not reliable predictors of the likelihood of ocular pathology in a child with headache, and the absence of such symptoms does not obviate the need for an ophthalmologist examination.

Disclosure statement
No potential conflict of interest was reported by the author(s).