Understanding the True Costs of NICU Care: Overlooked Variables That Could be Skewing Your Data
Infants admitted to a hospital’s Neonatal Intensive Care Unit (NICU) are understood to be those born prematurely or with medical complexities, warranting care that, by definition, is more intensive and also more costly than that provided to a healthy newborn. However, accepting the accuracy of this scenario may lead to a significant misinterpretation regarding your NICU population. Specifically, what happens to your analysis when the care of healthy newborns is being billed as NICU services?
As in any other industry, “you cannot manage what you cannot measure.” Health plans need to accurately assess the population they seek to manage and must start with a reliable data set to determine if and to what extent a problem exists. Within NICU populations, that analysis focuses on metrics such as level of care, length of stay, and cost per discharge as key indicators of care management efficiency.
This approach makes perfect sense – as long as the source data can provide a “perfect” perspective that can differentiate a “signal from noise”. When ProgenyHealth partners with health plans, our first step is to help them accurately identify and measure their NICU population. In our experience working with plans across the country for over 16 years, this perspective is seldom achieved because of a commonly overlooked data anomaly.
How the “Right” Data Analysis Can Produce the Wrong Assessment of NICU Population Health
Most health plan executives will look at their internal NICU population data and correctly focus on the KPIs of average length of stay (ALOS), cost per discharge (CPD), and percentage of days billed at NICU Levels 1-4. Comparing these metrics with similar managed plans, and seeing no major variances, they may conclude benchmarks are being hit, and there is no cause for concern. In our experience, this methodology is inaccurate and a good example of why achieving a clear, more refined view of population data is so important.
The potential problem arises if the plan is not isolating a commonly overlooked segment from the data set: healthy newborns included in the NICU population (based on the DRG) who may not have required NICU services. These newborns represent a relatively short stay and low-cost admission that likely may not meet medical necessity criteria for NICU services. The result: a skewed interpretation of the true LOS and CPD, and a significant barrier to effective population health management.
Diagnosing the NICU Data Anomaly
The cases that fall into this category are potentially avoidable NICU stays. These are generally newborns with normal birth weights and DRG assignments (e.g.: 640-1 and 640-2 or MS-DRG 795). Consistently, greater than 80% of total newborns in the data we review match this “well baby” criteria – the healthiest and lowest risk an infant can be. A NICU stay for such infants means that the health plan is paying for a level of service that may not meet medical necessity criteria for the NICU and, just as importantly, it causes dilution of utilization metrics within the NICU population. Assuming proper DRG assignment, NICU stays for these infants may not be warranted. But such stays do happen with enough frequency to mask broader utilization patterns.
Our recent analysis of DRG 640-1 and 640-2 length of stay metrics comparing a non-managed Medicaid plan with ProgenyHealth’s data revealed admissions billed at NICU DRGs that were well above Progeny-managed benchmarks for the shortest stays.
Note that the benchmarks shown above are only representative of this plan’s market, based on our analysis at ProgenyHealth. You should not compare your plan to this benchmark as yours may vary based on the population profile, regional factors, etc. With our nationwide data set and experience working with managed care plans for over 16 years, ProgenyHealth can help you arrive at appropriate and reliable benchmarks for your plan and project likely outcomes after our interventions.
Drilling down on these 1-3 Day ALOS cases, we identified a significant number of members with the lowest severity of illness (SOI), APR DRG 640-1 and 640-2, shown in the dotted line box below.
Again, assuming the DRG was grouped appropriately in all cases, these infants likely should not have been admitted to, or billed for, NICU services, and this plan did not have an accurate read on their population’s true NICU length of stay and associated costs. When the data set is adjusted to identify and extract these potentially avoidable NICU admissions, the before and after analysis provides a “true view” of the NICU utilization, as shown in the representative examples below.
Before (as-is NICU population)
After adjusting for “avoidable” NICU admissions
ALOS in Days
Avg. NICU Cost Per Discharge
This type of segmentation and analysis is critical for health plans to perform in order to assess where potential opportunities for improvement exist within their high-risk newborn population. NICU population health management by a team of seasoned specialists ensures every infant admitted to the NICU meets the medical necessity criteria for that hospitalization while also supporting accurate and timely payments to physicians and hospitals.