Study: Reporting of pediatric service provision varies significantly | Image Credit: © utah51 - © utah51 - stock.adobe.com.

A new cross-sectional study published in JAMA Network Open on June 3, 2025, revealed wide variability in how national datasets identify pediatric inpatient services at US-hospitals. The findings could have implications for how pediatric health care policy is informed and how access to care is measured.

Discrepancies across major hospital datasets

The analysis focused on 3 commonly used sources of hospital-level data: the American Hospital Association (AHA) Annual Survey, the Centers for Medicare & Medicaid Services Provider of Service (POS) file, and the National Pediatric Readiness Project (NPRP). Researchers compared the reporting of 4 pediatric inpatient services—newborn care, neonatal intensive care, general pediatric inpatient care, and pediatric intensive care—across 3,114 hospitals that appeared in all datasets in 2021. The NPRP was used as the benchmark for assessing reporting accuracy.

“Without validation of pediatric services as reported within national datasets, there is potential for misclassification of services, overestimation, and underestimation of service availability, and varied conclusions about hospital-level pediatric care,” the authors wrote. "Thus, we aimed to evaluate (1) the agreement between 3 national datasets in their reporting of 4 pediatric inpatient service lines (newborn, neonatal intensive care, general pediatrics, and pediatric intensive care), and (2) the precision and test characteristics of data reported associated with a pediatric benchmark."

According to the study, test characteristics and model performance were calculated and reported as F1 scores, "a machine learning evaluation metric that calculates the harmonic mean of precision and recall within a model," noted the investigative team.

Agreement was highest for newborn care, lowest for general pediatrics

Agreement between datasets varied substantially depending on the service. For newborn care, the AHA had the highest concordance with the NPRP, with an F1 score of 0.97 (95% CI, 0.96-0.97), compared to 0.62 (95% CI, 0.60-0.64) for POS. Neonatal intensive care services also showed relatively strong agreement in AHA data (F1 = 0.86; 95% CI, 0.85-0.88), while POS data performed moderately (F1 = 0.75; 95% CI, 0.74-0.77).

However, general pediatric inpatient care showed poorer alignment. AHA data had an F1 score of 0.69 (95% CI, 0.67-0.71), and POS data scored 0.79 (95% CI, 0.77-0.80). Pediatric intensive care services were inconsistently reported in POS (F1 = 0.49; 95% CI, 0.46-0.51) compared to stronger performance from AHA (F1 = 0.91; 95% CI, 0.90-0.93).

Implications for pediatric policy

The study highlighted how inconsistent data across widely-used national surveys may influence the apparent availability of pediatric services in the United States. As policymakers use these datasets to guide funding, emergency preparedness, and service planning, inaccurate or inconsistent service identification could perpetuate disparities in access to pediatric care.

The authors suggest that combining multiple data sources, while standardizing definitions and validating variables, may offer a more reliable path to capturing pediatric inpatient service availability nationwide.

“Accurate data that represent acute care hospital services for children are needed to inform policies that support access for children across the US,” the authors wrote. “Such data could enable policymakers to identify gaps in care, allocate resources more effectively, and ensure that children, regardless of their geographic location, receive the timely and specialized care they need.”

Reference:

McDaniel CE, Ralston Daniel M, Freyleue SD, et al. Identifying Inpatient Pediatric Services Across National Datasets. JAMA Netw Open. 2025;8(6):e2513527. doi:10.1001/jamanetworkopen.2025.13527