Why the Same Inpatient Claim Can Group to a Different MS-DRG

Coders reasonably expect a set of codes to have a DRG, the way a code has a title. It does not. A DRG is what a particular grouper package produced from those codes, and CMS revises the package.

CMS publishes a grouper per fiscal year and, for most years, a point release partway through it. The formula tables, code attributes, cluster tables, exclusions and hospital-acquired condition lists all belong to one specific package. Two dates of service that fall under different packages can classify the same clinical facts differently, and neither result is wrong.

The date of service, and nothing else. A grouping should never be produced from a clock, from whichever package happens to be newest, or from the nearest package to the date. A claim dated in a past year is answered by the grouper that governed it, or it is not answered at all.

That is also why a grouping result is only meaningful alongside its version. A DRG quoted without the package that produced it cannot be checked, and cannot be compared with another.

1. Record the fiscal year and the grouper package version alongside any DRG you report or compare.
2. When two systems disagree about a DRG, check the package versions before checking the codes. A version difference explains more disagreements than a coding difference does.
3. When a date of service is corrected, regroup. A date moved across a package boundary can move the result.
4. When reviewing which codes carry the DRG, regroup without each secondary diagnosis and each procedure in turn rather than reasoning about severity designations in the abstract.

A date correction. A claim is grouped, then the date of service is corrected by several months and crosses into a different grouper package. The codes have not changed, but the applicable formula tables have, and the claim must be regrouped. A result produced for the earlier date answers a question about a different set of inputs.

A code that holds the tier. Regrouping the claim without one secondary diagnosis drops it to a lower severity tier. That code is doing real work, and removing it during a review has a consequence a reviewer should see stated rather than infer.

A claim CMS returns. The grouper does not classify the claim at all and returns it with a code — an invalid principal diagnosis, an invalid discharge status, or a sex-specific code against an unstated sex. The reason is legible on the claim in front of you, and it is a coding problem rather than a grouping problem.

• Quoting a DRG without its grouper version. It cannot be checked or compared.
• Comparing groupings from different packages. The disagreement is often the version, not the coding.
• Assuming the newest grouper. An older claim is governed by the package in force on its date of service.
• Reading a grouping as a payment. A grouping is a classification of the stay. What a claim pays depends on hospital-specific rates and adjustments that a DRG does not carry.
• Treating an unavailable result as a clean one. Where the grouper tables for a fiscal year are not loaded, no DRG is shown at all, and an estimate presented as a grouping would answer a question nobody asked.

Grouping runs on the inputs a coder actually supplies: the principal diagnosis and its present-on-admission indicator, the secondary diagnoses each with their own indicator, the procedures, the sex and the discharge status. Nothing should be defaulted silently, because a defaulted input answers a different claim.

Present-on-admission reporting deserves particular care. It governs hospital-acquired condition handling, and an unreported or invalid indicator on a code that carries such a condition can change whether the claim groups at all.