Ten-year trends in infant neuroimaging from US Neonatal Intensive Care Units

An assessment of prevalence and expenditure associated with discharge brain MRI in preterm infants

To assess national expenditure associated with preterm-infant brain MRI and potential impact of reduction per Choosing Wisely campaign 2015 recommendation to “avoid routine screening term-equivalent or discharge brain MRIs in preterm-infants”. Cross-sectional U.S. trend data from the Agency for Healthcare Research and Quality (AHRQ), Healthcare Cost and Utilization Project (HCUP) Kids’ Inpatient Database (KID) database (2006, 2009, 2012, 2016) was used to estimate overall national expenditure associated with brain MRI among infants with gestational age (GA) ≤36 weeks, and also when classified as ‘not indicated’ (NI-MRI) i.e., equivalent to routine use without clinical indications and regarded as low-value service (LVS). Associated cost was determined by querying CMS-database for physician-fee-schedules to find the highest global procedure-cost per cycle, then adjusting for inflation. Sensitivity-analyses were conducted to account for additional clinical charges associated with NI-MRI. 3,768 (0.26%) of 1,472,236 preterm-infants had brain MRI across all cycles (inflation-adjusted total $3,690,088). Overall proportion of brain MRIs increased across 2006–2012 from 0.25%-0.33% but decreased in 2016 to 0.16% (P<0.001). Inflation-adjusted overall expenditure by cycle was: 2006, $1,299,130 (95% CI: $987,505, $1,610,755); 2009, $1,194,208 (95% CI: $873,487, $1,516,154); 2012, $931,836 (95% CI: $666,114, $1,197,156); and 2016, $264,648 (95% CI: $172,061, $357,280). Prevalence for NI-MRI in 2006, 2009, 2012 and 2016 was 86% (n = 809), 88% (n = 940), 89% (n = 1028) and 50% (n = 299), respectively; and 70% were in infants 35–36 weeks GA. NI-MRI prevalence was not different over time by payer-type (Medicaid, private), sex or race/ethnicity (white, black, Hispanic); larger hospital size was significantly associated across 2006–2012 but this declined for all sizes in 2016, with most decline in larger hospitals (P for interaction <0.05). NI-MRI expenditure sensitivity-analysis with addition of cycle median total-admission-charge to inflation-adjusted CMS-fee was $1,190,919/$518,343, for 2012/2016 cycles respectively. National MRI prevalence in preterm infants (both overall and LVS) and associated expenditure decreased substantially post recommendation; however, annual savings are modest and unlikely to be >$1.2 million.

The Denominator Matters! Lessons from Large Database Research in Neonatology

Observational studies from large datasets are becoming more common in neonatology. In this review, we highlight the importance of the denominator in study design and interpretation including examples of bias from source data, weight-based categories, age-related bias, and diagnosis-based denominators.