Factors Associated with Length of Stay and Cost among Pediatric Hospitalizations with a Primary Ophthalmic Diagnosis

ABSTRACT Purpose To investigate factors associated with prolonged length of stay and high cost among pediatric hospitalizations with a primary ophthalmic diagnosis. Methods This retrospective, cross-sectional study utilized data on pediatric admissions with a primary ophthalmic diagnosis from the multicenter 2016 Kids’ Inpatient Database. Multivariable logistic regression models adjusted for demographic, hospital, and admission characteristics were used to evaluate factors associated with prolonged stay and high cost, defined as exceeding the 75th percentile (>4 days and $12,642, respectively). Results An estimated 6,811 pediatric hospitalizations with a primary ophthalmic diagnosis in the United States in 2016 were included. On adjusted analysis, a prolonged length of stay was more likely with Medicaid (vs. private insurance, OR = 1.19, 95% CI: [1.02, 1.40], p = .03), non-trauma (vs. trauma, OR = 2.77, 95% CI: [2.12, 3.63], p < .001) and urban teaching hospitals (vs. rural, OR = 3.48, 95% CI: [1.04, 11.69], p = .04). A high cost of stay was more likely with higher income levels (Quartile 3 vs. 1, OR = 1.30, 95% CI: [1.02, 1.67], p = .04; Quartile 4 vs. 1, OR = 1.49, 95% CI: [1.08, 2.05], p = .02), private insurance (vs. Medicaid, OR = 1.26, 95% CI: [1.04, 1.53], p = .02), Western hospitals (vs. South, OR = 2.74, 95% CI: [1.83, 4.12], p < .001), and trauma (vs. non-trauma, OR = 3.29, 95% CI: [2.57, 4.21], p < .001). Children and young adults had higher odds of prolonged stay, while adolescents and young adults had higher odds of high cost compared to toddlers (p < .05 for all). Conclusions Additional work addressing the factors associated with higher resource utilization may help promote the delivery of quality inpatient pediatric eye care.


Introduction
Despite primarily being an outpatient specialty, there are indications for inpatient care within the realm of ophthalmology. 1 Such care is needed to treat and manage potentially vision-threatening or systemic conditions. From 2001 to 2014, it was estimated that 16 in 100,000 persons required hospitalization for a primary ophthalmic diagnosis in the United States (U.S.), associated with an annual cost of $422 million/year. 1 Pediatric hospitalizations made up nearly 25% of these admissions. 1 Hospital costs comprise the largest portion of pediatric healthcare spending. 2 Not only are pediatric hospital costs increasing, 3 but there is evidence that ophthalmology is among the specialties with the greatest growth in hospital costs for children. 4 Outcomes such as length of stay and costs are important measures of resource utilization that healthcare systems seek to improve. 5 Understanding factors associated with high resource utilization is an important first step to identifying areas of improvement and moving towards value-based care. Prior work in inpatient ophthalmic care has demonstrated that outcomes such as length of stay and cost may vary based on demographic and admission characteristics. 6,7 However, such work has yet to be done for children primarily hospitalized for an ophthalmic condition. Given the resource burden of pediatric inpatient care, more investigation is needed to understand factors associated with high hospital resource utilization among pediatric admissions with a primary ophthalmic diagnosis.
The objective of this study is to identify risk factors for prolonged length of stay and high cost of stay for pediatric admissions with a primary diagnosis of an ophthalmic condition.

Data source
We conducted a cross-sectional retrospective analysis of the 2016 Kids' Inpatient Database (KID), produced by the Healthcare Cost and Utilization Project (HCUP) under the Agency for Healthcare Research and Quality. The KID is the largest all-payer database designed to produce national estimates of pediatric inpatient stays for patients less than 21 years old, using a nationwide sample of pediatric inpatient discharges in the US, and is produced in three-year intervals. 8 The 2016 KID version includes 4,200 community, non-rehabilitation hospitals from 46 states and the District of Columbia, sampling 10% of normal newborns and 80% of pediatric discharges from these hospitals. The resulting 3.1 million pediatric discharges represent a target of 5,001 hospitals and nearly 6.3 million pediatric discharges in 2016. Data were extracted from all discharges in the KID 2016 dataset with a principal diagnosis of an ophthalmic condition, using International Classification of Diseases-10 th Edition-Clinical Modification (ICD-10) codes consistent with prior studies 1,9 (Table 1). This project received approval from the Johns Hopkins School of Medicine Institutional Review Board and abides by the guidelines of the Declarations of Helsinki.

Statistical analysis
To produce national estimates, data were weighted using discharge weights provided by the KID. All analysis was conducted using survey methods that account for the sampling method and distribution weights used in the KID. 8 Outcomes assessed included prolonged length of stay and high cost of stay, both of which were defined as exceeding the 75 th percentile of the distribution in this sample. The 75 th percentile has been used to determine a cut-off for high versus low values in prior studies using HCUP data including the KID. 6,10 Costs were calculated from hospital charges using hospital-specific cost-tocharge ratios provided by KID and inflation-adjusted to 2020-dollar values using the Consumer Price Index for Hospital Services from the US Bureau of Labor Statistics. 11 Demographic and hospital characteristics included sex, age, race, income quartile based on patient ZIP code, primary payer, hospital region, and hospital location-teaching status. Age was categorized into infants (<1 year), toddlers (1-<6 years), children (6-<13 years), adolescents (13-<18 years), and young adults (18-<21 years). Admission characteristics included eye trauma status and overall severity. Eye trauma status as a dichotomous variable was assigned based on the principal diagnosis (Table 1). In the KID, severity (minor, moderate, major, extreme) is assigned to each hospital stay using All-Patient Refined Diagnosis Related Groups, which accounts for variables such as principal diagnosis, age, interaction of secondary diagnoses, and procedures performed. 12 Patients with missing data for any of the outcomes, demographic, and hospital variables were excluded (unweighted n = 744). Descriptive analysis was conducted for all demographic, hospital, and admission characteristics. Univariable and multivariable logistic regression models were used to examine the association between each characteristic and prolonged length of stay or high cost of stay. All demographic, hospital, and admission covariates were included in the multivariable models based on clinical relevance, with the inclusion of length of stay as an additional covariate in the cost model. All analyses were conducted in Stata SE/17.0 (Stata Corp LP, College Station, TX). Statistical significance was defined at p < .05.

Demographic and admission characteristics
This analysis included an estimated 6,811 pediatric admissions across the country with a primary ophthalmic diagnosis in 2016 ( Table 2). The majority of admissions were male (N = 3,913, 57.4%) and nearly half were White (N = 3,344, 49.1%). The average age was 8.2 years old (SD = 6.5 years). Most patients had Medicaid (N = 3,365, 49.4%) and were from households in the lowest income quartile (N = 2,200, 32.3%). The majority

Discussion
This study used a nationally representative pediatric database to understand disparities in resource utilization among pediatric admissions owing to a primary ophthalmic diagnosis, namely length of stay and cost. Key factors associated with prolonged length of stay and high cost among pediatric ophthalmic admissions included income level, insurance type, hospital region, hospital teaching status, trauma status, and age.

Income level and insurance
In our study, patients with public insurance had higher odds of having a prolonged length of stay. Those with higher income levels or with private insurance were more likely to have a high cost of stay. With regards to length of stay, prior studies using the KID have similarly found that public insurance is associated with longer stays among various indications for pediatric hospitalizations. 10,13,14 Additionally, the relationship between private insurance and higher hospital costs is consistent with data on non-birth pediatric admissions nationally in 2016 15 and has been demonstrated in prior work investigating costs in admissions for ocular trauma. 6 There may be differences in care received by income level and insurance status contributing to disparate lengths of stay and costs found in our study. For example, prior studies have found that relative to private insurance, pediatric patients with public insurance receive less diagnostic testing and intervention 16 but have a greater likelihood of being kept for observation or admission. 17 To that end, further studies are needed to first identify whether the differences in length of stay and costs between insurance type and income levels found among pediatric ophthalmic admissions in our study represent under-or overutilization of hospital resources and for which patients. Additional research is needed to investigate practice patterns that may contribute to the differences in resource outcomes between insurance types and income levels and their root causes, whether it be provider bias or larger institutional factors.

Teaching hospital
Admissions in urban teaching hospitals were more likely to have prolonged length of stays relative to rural nonteaching hospitals, even after adjusting for severity to account for potential differences in the complexity of cases seen between teaching and non-teaching hospitals. Longer lengths of stay among pediatric admissions in teaching vs. non-teaching hospitals have been demonstrated in other studies 18,19 ; the additional academic activities performed and different levels of trainees providing care in teaching hospitals could be contributing factors. 20 As over 90% of admissions in our study occurred at urban teaching hospitals, taking measures to address the length of stay in this setting can have a large impact on resource utilization across pediatric ophthalmic admissions.

Hospital region
Consistent with prior research on hospital costs in ocular trauma 6 and other causes of pediatric hospitalizations, 21,22 admissions to hospitals in the West were more likely to incur a high cost of stay.
Review of national data from 2016 indicates that the high inpatient cost of Western hospitals may be primarily driven by states belonging to the Pacific census division (California, Washington, Alaska, and Hawaii). 23 These regional differences likely reflect differences in local market pricing, inflation rates, cost of living, and other factors; however, it is unknown whether there are regional variations in hospital practices that may contribute to higher costs in the West.

Trauma
After adjusting for severity, traumatic admissions were more likely to have a high cost of stay but less likely to have a prolonged length of stay. The association with higher costs may be expected given the need for procedures in trauma, while the decreased likelihood of a prolonged stay may be explained by the more acute nature of the interventions in traumatic versus non- traumatic admissions. One study found that among children with subperiosteal abscesses of the orbit and similar disease burdens, medical management resulted in longer hospital stays than surgical management. 24 Likewise, surgical management in traumatic causes of ophthalmic admission may provide quicker, definitive treatment compared to medical causes of admission that may require prolonged periods of therapy and observation (e.g., antibiotic therapy) to achieve resolution.

Age
Using toddlers (1-5 years) as the reference group, children (6-12 years) and young adults (18-20 years) had higher odds of prolonged stay. Adolescents and young adults also had higher odds of a high cost of stay. While our analyses adjusted for overall severity and accounted for traumatic vs. non-traumatic admissions, the profile of diagnoses within the larger trauma or non-trauma categories may differ by age group and impact expected length of stay and cost between these age groups. For example, orbital cellulitis and eyelid abscesses make up a much larger portion of non-trauma ophthalmic admissions in the toddler category than in any other age category; trauma in toddlers most often consists of ocular laceration while orbital floor fractures become much more prevalent in older age categories. 1 Thus, differences in resource utilization detected in our study between toddlers and pediatric patients of other ages may be owing to the different conditions commonly treated in these age groups.

Limitations
This study is subject to several limitations inherent to administrative databases. While this is a nationally representative dataset, the data are limited to only recorded discharges. Additionally, as each discharge is an observation, a patient may be counted more than once if admitted more than once. Other limitations include missing data or errors in coding or billing. Costs are converted from charges billed for the entire hospital stay and do not include professional (physician) fees. While this study used ICD-10 codes based on ICD-9 codes included in prior research regarding hospitalizations with a primary ophthalmic diagnosis, future research may build upon this work by including additional eye-related ICD-10 codes, as well as by analyzing the costs and length of stays among specific eye conditions.

Conclusion
Inpatient care comprises a majority of healthcare costs for children and continues to be increasingly so. Across all specialties, including in ophthalmology, it is important to examine ways to reduce resource burden. Using a nationally representative dataset, this study identified several disparities in resource utilization among children hospitalized for a primary ophthalmic condition. Understanding and addressing the mechanisms behind these disparities may help promote the delivery of quality eye care for children.