# ICE-T Inpatient Cost Evaluation Tool

# Using ICE–T to Make a Bundle

There are many factors to consider when determining which inpatient DRG to bundle and exactly how radiology practices will participate in the bundled payment.

Perhaps the most important factors to consider are the three “V”s: **V**alue, **V**ariance, and **V**olume

ICE-T provides information on the three Vs by aggregating many years of national inpatient Medicare data by DRG and presenting it in an easy to use graphical interface through the Compare tool.

Although all three Vs are important by themselves, it is the interaction between value, variance, and volume that determine which DRGs are best for a practice to bundle and the appropriate risk-mitigating price.

**Interaction between Value, Variance, and Volume**

ICE-T uses the mean cost of an inpatient DRG to estimate the **value** of bundling the DRG. When a DRG has a low mean imaging cost, it may be difficult for practices to find savings – making it unprofitable to participate in a bundle for that DRG, unless there is very little variation in episode costs. High value DRGs, however, may provide more opportunities for savings but can expose practices to larger losses if they face too many high cost episodes. A DRG’s value must always be considered in the context of the variance in episode cost for that DRG.

**Variance** measures how dispersed the episode costs in a DRG are from the mean cost of the episode. Understanding the variance in the cost of a DRG is important to understanding if a bundled payment price will cover the expected risk. Higher variance in the DRG cost suggests that bundled payments will not cover the costs of these claims. Variance in episode costs is depicted graphically in the Compare tool when the interquartile box (which spans the difference between the 25th to 75th quartiles of episode cost) is longer. Because inpatient health care costs are often heavily right-skewed, DRGs where the mean cost is closer to the 50th quartile have less variance – and thus less risk – than DRGs where the mean cost is closer to the 75th quartile. This measure of variance captures the influence of the outliers that extend beyond the 75th quartile.

When developing bundled payments, neither value nor variance matter if there is not enough episode **volume **to adequately average out the episode’s costs in order to cover the risk of high cost episodes. The frequency reported by ICE-T in the Compare tool is the number of episodes for a given DRG in our data. It is easier to develop a bundle for a DRG with higher frequencies as providers are more likely to face the true average cost for an episode. However, these are the Part A inpatient DRG volumes associated with a 5% sample of Medicare enrollees and are best used as *relative* comparisons between different DRGs rather than as *absolute* estimates of DRG volume.

You can learn more about the statistical terms we use in the Glossary. It is also important to consider how different the minimum and maximum cost values are from the interquartile box, as this will influence the cost-effectiveness of a bundled payment.

**Case Studies for Using and Interpreting ICE-T**

Below are fictional case studies designed to demonstrate the bundling process for radiology practices and how ICE-T can help.

**Case 1: Which DRG Should a Practice Bundle?**

**Case 2: Examining CJR Episodes**