Sociodemographic predictors of imaging utilization in children with right lower quadrant pain.

Abstract

Background: Inequities in health care access lead to inequities in outcome. In recent years, health outcome disparities have been documented in children with acute appendicitis and sociodemographic predictors of imaging utilization have not been adequately assessed.

Objective: The purpose of our study is to assess sociodemographic predictors for the diagnostic imaging of children with right lower quadrant (RLQ) pain. Our hypothesis is that disparities exist in imaging utilization.

Materials and methods: Our nationwide retrospective cohort study of the Pediatric Health Information System (PHIS) database queried emergency department encounters for children aged 0-18 years presenting with RLQ pain (ICD code CM R10.31) between January 2018 and September 2023. Primary exposures included neighborhood-level sociodemographic metrics as measured by Child Opportunity Index, race/ethnicity, and insurance status. Outcomes included no diagnostic imaging, diagnostic imaging with radiography alone, ultrasonography (US), computed tomography (CT), and magnetic resonance imaging (MRI). Multivariable logistic regression analyses assessed modality usage with respect to the primary exposures after controlling for demographic (age, gender) and additional (hospital geographic region, time of imaging) covariates. To avoid the perpetuation of bias, reference categories were determined by the lowest numerical value for each covariate.

Results: In total, 100,161 patient encounters met inclusion criteria (mean patient age 11.2 years ± 3.9; 59.3%, n = 59,416 females). Imaging utilized was US (78.0%; n = 78,115), CT (16.4%, n = 16,405), no imaging (13.9%, n = 13,894), radiography alone (4.4%, n = 4,429), and MRI (3.1%, n = 3,148). The most predictive factors for no imaging were moderate, low, and very low Child Opportunity Index (aOR 1.25, 1.17, and 1.18 [95% CI 1.10-1.33] compared to very high Child Opportunity Index); Black race/ethnicity (aOR 1.26 [95% CI 1.11-1.44] compared to White or Asian race/ethnicity); and public or other insurance (aOR 1.23 and 1.32 [95% CI 1.18-1.41] compared to commercial insurance). The most predictive factors for radiography alone were Black race/ethnicity (aOR 1.30 [95% CI 1.17-1.45] compared to Hispanic race/ethnicity) and public or other insurance (aOR 1.26 [95% CI 1.11-1.44] compared to commercial). The most predictive factors for US were very-high Child Opportunity Index (aOR 1.16 [95% CI 1.09-1.22] compared to very low Child Opportunity Index); Asian, NH-White, or Hispanic race/ethnicity (aOR 1.33, 1.31, 1.30 [95% CI 1.18-1.40] compared to Black race/ethnicity); and commercial insurance (aOR 1.20 [95% CI 1.16-1.25] compared to public insurance). The most predictive factor for CT was White race/ethnicity (aOR 1.26 [95% CI 1.11-1.43] compared with Asian race/ethnicity) and the most predictive factor for MRI was Hispanic race/ethnicity (aOR 1.49 [95% CI 1.17-1.61] compared with Black race/ethnicity). The greatest predictor of cross-sectional imaging was a hospital's region, with CT most likely in southern hospitals (aOR 4.09 [95% CI 2.17-7.70] compared with northeast hospitals). Patient Child Opportunity Index did not predict the likelihood of cross-sectional imaging with CT or MRI in tertiary pediatric centers.

Conclusion: Sociodemographic disparities exist in the imaging of children presenting to tertiary pediatric hospitals with RLQ pain. Future studies are needed to analyze the causes of such disparities both on hospital and departmental levels.