LAUSR

Importance Infants who appear neurologically well and have fractures concerning for abuse are at increased risk for clinically occult head injuries. Evidence of excess variation in neuroimaging practices when abuse is suspected may indicate opportunity for quality and safety improvement. Objective To quantify neuroimaging practice variation across children's hospitals among infants with fractures evaluated for abuse, with the hypothesis that hospitals would vary substantially in neuroimaging practices. As a secondary objective, factors associated with neuroimaging use were identified, with the hypothesis that age and factors associated with potential biases (ie, payer type and race or ethnicity) would be associated with neuroimaging use. Design, Setting, and Participants This cross-sectional study included infants with a femur or humerus fracture or both undergoing abuse evaluation at 44 select US children's hospitals in the Pediatric Health Information System (PHIS) from January 1, 2016, through March 30, 2020, including emergency department, observational, and inpatient encounters. Included infants were aged younger than 12 months with a femur or humerus fracture or both without overt signs or symptoms of head injury for whom a skeletal survey was performed. To focus on infants at increased risk for clinically occult head injuries, infants with billing codes suggestive of overt neurologic signs or symptoms were excluded. Multivariable logistic regression was used to investigate demographic, clinical, and temporal factors associated with use of neuroimaging. Marginal standardization was used to report adjusted percentages of infants undergoing neuroimaging by hospital and payer type. Data were analyzed from March 2021 through January 2022. Exposures Covariates included age, sex, race and ethnicity, payer type, fracture type, presentation year, and hospital. Main Outcomes and Measures Use of neuroimaging by CT or MRI. Results Of 2585 infants with humerus or femur fracture or both undergoing evaluations for possible child abuse, there were 1408 (54.5%) male infants, 1726 infants (66.8%) who were publicly insured, and 1549 infants (59.9%) who underwent neuroimaging. The median (IQR) age was 6.1 (3.2-8.3) months. There were 748 (28.9%) Black non-Hispanic infants, 426 (16.5%) Hispanic infants, 1148 (44.4%) White non-Hispanic infants. In multivariable analyses, younger age (eg, odds ratio [OR] for ages <3 months vs ages 9 to <12 months, 13.2; 95% CI, 9.54-18.2; P < .001), male sex (OR, 1.47; 95% CI, 1.22-1.78; P < .001), payer type (OR for public vs private insurance, 1.48; 95% CI, 1.18-1.85; P = .003), fracture type (OR for femur and humerus fracture vs isolated femur fracture, 5.36; 95% CI, 2.11-13.6; P = .002), and hospital (adjusted range in use of neuroimaging, 37.4% [95% CI 21.4%-53.5%] to 83.6% [95% CI 69.6%-97.5%]; P < .001) were associated with increased use of neuroimaging, but race and ethnicity were not. Publicly insured infants were more likely to undergo neuroimaging (62.0%; 95% CI, 60.0%-64.1%) than privately insured infants (55.1%; 95% CI, 51.8%-58.4%) (P = .001). Conclusions and Relevance This study found that hospitals varied in neuroimaging practices among infants with concern for abuse. Apparent disparities in practice associated with insurance type suggest opportunities for quality, safety, and equity improvement.