Data Source And Extraction

This cross-sectional study of a large commercial insurance claims database was deemed exempt from review by the institutional review board at the University of California Los Angeles, and all procedures were in accordance with the tenets of the Declaration of Helsinki. Data was derived from OLDW, a database that includes de-identified, longitudinal health information on enrollees and patients with a broad range of age, race and ethnicity, and geographical distribution across the United States. With claims data from commercial insurance and Medicare Advantage enrollees, excluding Medicare Part B (Fee for Service) and medical assistance programs, the claims data in OLDW is inclusive of medical and pharmacy claims, laboratory results, and enrollment records.

Medical claims data between January 1, 2007, and December 31, 2018, were extracted from the OLDW for all subjects with medical coverage during those dates. Given that the purpose of this study was to increase the breadth and depth of the understanding of the association of serious structural eye disease and mental illness, the most common serious eye conditions that presented in this database, as published previously, were included. Patients with eye disease (glaucoma, cataract, congenital optic nerve disease, congenital retinal disease, and blindness/low vision due to all causes) or mental illness (anxiety disorder, schizophrenia, bipolar disorder, depressive disorder, and substance use disorder) were identified based on diagnosis codes in the International Classification of Disease, Ninth Revision, Clinical Modification (ICD-9-CM) or the International Classification of Disease, Tenth Revision, Clinical Modification (ICD-10-CM) as listed in Supplemental Table 1. The child’s age was calculated at the date of first claim for an eye disease. The age at first claim of each subtype of eye disease was also calculated. Controls consisted of children and teens without any eye disease diagnoses other than refractive error. Their age was calculated at the date of their first medical claim (between January 1, 2007 and December 31, 2018). Subjects were excluded from further analysis if they met the following criteria: those who were more than 18 years of age at the corresponding first claim; those with the corresponding first claims after January 1, 2018; 3); those without medical insurance; those with less than 6 months of medical insurance since their corresponding first claim; and those with eye diseases other than the 5 specific eye diseases mentioned.

In addition to age, baseline demographics including sex, race and ethnicity, education level of parents, household income, family net worth, geographic region, and type of insurance were also extracted. As a measure of general health, the presence or absence of medical comorbidities was also extracted, using a subset of the Agency for Healthcare Research and Quality (AHRQ) comorbidity variables selected based on their inclusion in the Pediatric Comorbidity Index, a model used to predict 1-year mortality in pediatric patients. These variables include septicemia, mycoses, cancer of the brain and nervous system, leukemias, coagulation and hemorrhagic disorders, diseases of white blood cells, epilepsy/convulsions, essential hypertension, hypertension with complications and secondary hypertension, conduction disorders, cardiac dysrhythmias, cardiac arrest and ventricular fibrillation, congestive heart failure/nonhypertensive, acute cerebrovascular disease, pneumonia, influenza, aspiration pneumonitis/food/vomitus, respiratory failure/insufficiency/arrest, nervous system congenital anomalies, intrauterine hypoxia and birth asphyxia, fracture of neck of femur, intracranial injury, fever of unknown origin, shock, and developmental disorders.

STATISTICAL ANALYSIS

Data extractions were conducted using SQL Software DBVisualizer Pro 10.0.15 (DbVis Software AB), and statistical analyses were performed using R (3.5.3) (R Foundation for Statistical Computing). Descriptive statistics were performed for children and teens with and without eye diseases. Continuous variables were compared using t tests, and categorical variables were compared using χ ² tests. The unadjusted odds ratio for mental illness on eye disease was calculated with 95% confidence intervals using univariate logistic regression models. Adjusted effects were estimated using multivariable logistic regression models with potential confounding variables, including age, sex, race and ethnicity, education, family net worth, geographic region, and the presence of at least 1 of the comorbidities as an indication for the status of general health.

To examine the effect of eye disease on each mental illness, children and teens with other types of mental illnesses were excluded from the corresponding analysis, and children and teens without any mental illnesses served as the reference group in all the analyses. Similarly, when assessing the unique effect of each subtype of eye disease, children and teens with other subtypes of eye disease were further excluded so that those without any eye diseases were kept in the control group in all comparisons.