Spending Variation in Medicaid Newborn Care

About half of births in the US are paid by Medicaid, and newborns account for a high percentage of the Medicaid population. While regional variation in the care of privately and Medicare insured populations have been widely studied, less is known about Medicaid populations. We measure the contribution of price, quantity and health risk to regional spending variation in Medicaid newborn care.

The study uses Texas Medicaid claims, encounters and enrollment files linked to birth and death certificates for Medicaid newborns born between January 1, 2014 and December 31, 2014 with birth weight of at least 400 grams. To examine spending variation, we define 21 neonatal intensive care regions (NICR) that reflect geographic markets for newborn medical care using small area analysis methods.

Spending is computed as spending per newborn hospital stay including between hospital transfers until first discharge to home or death, and is categorized by service category into inpatient facility and related professional spending. We follow the method by White (2012) and Franzini et al (2014) to decompose 1) the spending index into a price index and a quantity index, 2) the price index into an input price index and an adjusted price, and 3) the quantity index into a health risk index and an adjusted quantity index. We compute the spending index by comparing NICR mean newborn care spending to those of Texas by service category. To exam whether the price is higher or lower than state average, we compare actual spending in NICRs to hypothetical spending estimated using standard prices. The standard price is calculated using Texas mean spending for each APR-DRG in inpatient facility spending and for each CPT/HCPCS code in professional spending. The price index is then adjusted for the local input price using Medicare wage index for facility spending and Medicare Geographical adjustment factors (GAF) for professional spending. The quantity index is computed by dividing the spending index by the price index. Since neonatal health risk differs across regions, quantity index is adjusted to account for these differences. Then, we compute the coefficients of variation (CV, standard deviation divided by the mean) to measure variation by NICR for spending per newborn stay and allocate the share of spending variation attributable to each component of variations-input price, adjusted price, health status and adjusted quantity-using the weighted variance-covariance matrix of the natural logarithm of the indices.

We find that significant regional spending variation exists in Medicaid newborn care across Texas. It is driven by differences in prices and utilization as well as differences in health risk.